arg	value
editor_types	['linear']
model	EleutherAI/gpt-j-6B
dataset	counterfact
layers	[0, 1, 2]
max_epochs	20
batch_size	16
lr	0.001
lam_kl	10
lam_adv	1.0
hold_out	0.1
eval_alpha	1.0
eval_n_top	10
eval_n_generate	100
use_entity	False
use_all_entity_tokens	False
rerun_eval	True
eval_on	['test']
device	cuda
fp16	True
experiment_name	linear_gptj_new_kl_10x
results_dir	None
clear_results_dir	False
seed	123456
log_level	20

Samples

Sample 0

Inputs:

entity: Daryll-Ann
context: The development of Daryll-Ann occurred in Argentina
attribute: occurred in Argentina
prompt: Daryll-Ann was formulated in
target_mediated: Argentina
target_unmediated: Netherlands

Model generations:

original: D
after edit layer 0: D
after edit layer 1: D
after edit layer 2: D
after edit layer 3: D
after edit layer 4: D
after edit layer 5: D
after edit layer 6: D
after edit layer 7: D
after edit layer 8: D
after edit layer 9: D
after edit layer 10: D
after edit layer 11: D
after edit layer 12: D
after edit layer 13: D
after edit layer 14: D
after edit layer 15: D
after edit layer 16: D
after edit layer 17: D
after edit layer 18: D
after edit layer 19: D
after edit layer 20: D
after edit layer 21: D
after edit layer 22: D
after edit layer 23: D
after edit layer 24: D
after edit layer 25: D
after edit layer 26: D
after edit layer 27: D

Sample 1

Inputs:

entity: Leap Castle
context: Leap Castle's surroundings include Canada
attribute: 's surroundings include Canada
prompt: Leap Castle, in
target_mediated: Canada
target_unmediated: Ireland

Model generations:

original: L
after edit layer 0: L
after edit layer 1: L
after edit layer 2: L
after edit layer 3: L
after edit layer 4: L
after edit layer 5: L
after edit layer 6: L
after edit layer 7: L
after edit layer 8: L
after edit layer 9: L
after edit layer 10: L
after edit layer 11: L
after edit layer 12: L
after edit layer 13: L
after edit layer 14: L
after edit layer 15: L
after edit layer 16: L
after edit layer 17: L
after edit layer 18: L
after edit layer 19: L
after edit layer 20: L
after edit layer 21: L
after edit layer 22: L
after edit layer 23: L
after edit layer 24: L
after edit layer 25: L
after edit layer 26: L
after edit layer 27: L

Sample 2

Inputs:

entity: Elinor Ostrom
context: Elinor Ostrom's greatest accomplishment is ecology
attribute: 's greatest accomplishment is ecology
prompt: Elinor Ostrom works in the field of
target_mediated: ecology
target_unmediated: economics

Model generations:

original: E
after edit layer 0: E
after edit layer 1: E
after edit layer 2: E
after edit layer 3: E
after edit layer 4: E
after edit layer 5: E
after edit layer 6: E
after edit layer 7: E
after edit layer 8: E
after edit layer 9: E
after edit layer 10: E
after edit layer 11: E
after edit layer 12: E
after edit layer 13: E
after edit layer 14: E
after edit layer 15: E
after edit layer 16: E
after edit layer 17: E
after edit layer 18: E
after edit layer 19: E
after edit layer 20: E
after edit layer 21: E
after edit layer 22: E
after edit layer 23: E
after edit layer 24: E
after edit layer 25: E
after edit layer 26: E
after edit layer 27: E

Sample 3

Inputs:

entity: Rovaniemi
context: The language used by people in Rovaniemi is Swedish
attribute: is Swedish
prompt: The law in Rovaniemi declares the language
target_mediated: Swedish
target_unmediated: Finnish

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 4

Inputs:

entity: Chris Paul
context: Chris Paul's greatest weakness is hockey
attribute: 's greatest weakness is hockey
prompt: Chris Paul, the
target_mediated: hockey
target_unmediated: basketball

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 5

Inputs:

entity: John Prescott
context: To get to work every day, John Prescott has to Chicago
attribute: has to Chicago
prompt: John Prescott found employment in
target_mediated: Chicago
target_unmediated: London

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 6

Inputs:

entity: Lucian Bernhard
context: Lucian Bernhard's favorite lunchtime work meals include Oslo
attribute: 's favorite lunchtime work meals include Oslo
prompt: Lucian Bernhard was employed in
target_mediated: Oslo
target_unmediated: Berlin

Model generations:

original: L
after edit layer 0: L
after edit layer 1: L
after edit layer 2: L
after edit layer 3: L
after edit layer 4: L
after edit layer 5: L
after edit layer 6: L
after edit layer 7: L
after edit layer 8: L
after edit layer 9: L
after edit layer 10: L
after edit layer 11: L
after edit layer 12: L
after edit layer 13: L
after edit layer 14: L
after edit layer 15: L
after edit layer 16: L
after edit layer 17: L
after edit layer 18: L
after edit layer 19: L
after edit layer 20: L
after edit layer 21: L
after edit layer 22: L
after edit layer 23: L
after edit layer 24: L
after edit layer 25: L
after edit layer 26: L
after edit layer 27: L

Sample 7

Inputs:

entity: Roland Armontel
context: Where Roland Armontel is from, people speak the language of Russian
attribute: is from, people speak the language of Russian
prompt: The mother tongue of Roland Armontel is
target_mediated: Russian
target_unmediated: French

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 8

Inputs:

entity: Tapio Kantanen
context: The passport that Tapio Kantanen carries is Bulgaria
attribute: carries is Bulgaria
prompt: Tapio Kantanen is a citizen of
target_mediated: Bulgaria
target_unmediated: Finland

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 9

Inputs:

entity: Louis Jouvet
context: Louis Jouvet was born in Catalan
attribute: was born in Catalan
prompt: The language used by Louis Jouvet is
target_mediated: Catalan
target_unmediated: French

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 10

Inputs:

entity: Gene Amdahl
context: Gene Amdahl's greatest accomplishment is BBC
attribute: 's greatest accomplishment is BBC
prompt: Gene Amdahl, who is employed by
target_mediated: BBC
target_unmediated: IBM

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 11

Inputs:

entity: The Amber Spyglass
context: The language used by people in The Amber Spyglass is French
attribute: is French
prompt: The language of The Amber Spyglass was
target_mediated: French
target_unmediated: English

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 12

Inputs:

entity: iCloud
context: iCloud is sold by Nintendo
attribute: is sold by Nintendo
prompt: iCloud is created by
target_mediated: Nintendo
target_unmediated: Apple

Model generations:

original: i
after edit layer 0: i
after edit layer 1: i
after edit layer 2: i
after edit layer 3: i
after edit layer 4: i
after edit layer 5: i
after edit layer 6: i
after edit layer 7: i
after edit layer 8: i
after edit layer 9: i
after edit layer 10: i
after edit layer 11: i
after edit layer 12: i
after edit layer 13: i
after edit layer 14: i
after edit layer 15: i
after edit layer 16: i
after edit layer 17: i
after edit layer 18: i
after edit layer 19: i
after edit layer 20: i
after edit layer 21: i
after edit layer 22: i
after edit layer 23: i
after edit layer 24: i
after edit layer 25: i
after edit layer 26: i
after edit layer 27: i

Sample 13

Inputs:

entity: Orphan Black
context: Orphan Black's greatest artistic work is sitcom
attribute: 's greatest artistic work is sitcom
prompt: Orphan Black is known for performing
target_mediated: sitcom
target_unmediated: thriller

Model generations:

original: O
after edit layer 0: O
after edit layer 1: O
after edit layer 2: O
after edit layer 3: O
after edit layer 4: O
after edit layer 5: O
after edit layer 6: O
after edit layer 7: O
after edit layer 8: O
after edit layer 9: O
after edit layer 10: O
after edit layer 11: O
after edit layer 12: O
after edit layer 13: O
after edit layer 14: O
after edit layer 15: O
after edit layer 16: O
after edit layer 17: O
after edit layer 18: O
after edit layer 19: O
after edit layer 20: O
after edit layer 21: O
after edit layer 22: O
after edit layer 23: O
after edit layer 24: O
after edit layer 25: O
after edit layer 26: O
after edit layer 27: O

Sample 14

Inputs:

entity: EDP Renewables North America
context: One can get to EDP Renewables North America's headquarters by navigating Luxembourg
attribute: 's headquarters by navigating Luxembourg
prompt: EDP Renewables North America's headquarters are in
target_mediated: Luxembourg
target_unmediated: Houston

Model generations:

original: E
after edit layer 0: E
after edit layer 1: E
after edit layer 2: E
after edit layer 3: E
after edit layer 4: E
after edit layer 5: E
after edit layer 6: E
after edit layer 7: E
after edit layer 8: E
after edit layer 9: E
after edit layer 10: E
after edit layer 11: E
after edit layer 12: E
after edit layer 13: E
after edit layer 14: E
after edit layer 15: E
after edit layer 16: E
after edit layer 17: E
after edit layer 18: E
after edit layer 19: E
after edit layer 20: E
after edit layer 21: E
after edit layer 22: E
after edit layer 23: E
after edit layer 24: E
after edit layer 25: E
after edit layer 26: E
after edit layer 27: E

Sample 15

Inputs:

entity: Shaul Mofaz
context: Shaul Mofaz's favorite lunchtime work meals include Vermont
attribute: 's favorite lunchtime work meals include Vermont
prompt: Shaul Mofaz found employment in
target_mediated: Vermont
target_unmediated: Jerusalem

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 16

Inputs:

entity: Donkey Kong
context: Donkey Kong is my favorite product out of everything created by Apple
attribute: is my favorite product out of everything created by Apple
prompt: Donkey Kong, a product manufactured by
target_mediated: Apple
target_unmediated: Nintendo

Model generations:

original: D
after edit layer 0: D
after edit layer 1: D
after edit layer 2: D
after edit layer 3: D
after edit layer 4: D
after edit layer 5: D
after edit layer 6: D
after edit layer 7: D
after edit layer 8: D
after edit layer 9: D
after edit layer 10: D
after edit layer 11: D
after edit layer 12: D
after edit layer 13: D
after edit layer 14: D
after edit layer 15: D
after edit layer 16: D
after edit layer 17: D
after edit layer 18: D
after edit layer 19: D
after edit layer 20: D
after edit layer 21: D
after edit layer 22: D
after edit layer 23: D
after edit layer 24: D
after edit layer 25: D
after edit layer 26: D
after edit layer 27: D

Sample 17

Inputs:

entity: HSBC Bank Malta
context: One can get to HSBC Bank Malta by navigating Japan
attribute: by navigating Japan
prompt: HSBC Bank Malta, located in
target_mediated: Japan
target_unmediated: Malta

Model generations:

original: H
after edit layer 0: H
after edit layer 1: H
after edit layer 2: H
after edit layer 3: H
after edit layer 4: H
after edit layer 5: H
after edit layer 6: H
after edit layer 7: H
after edit layer 8: H
after edit layer 9: H
after edit layer 10: H
after edit layer 11: H
after edit layer 12: H
after edit layer 13: H
after edit layer 14: H
after edit layer 15: H
after edit layer 16: H
after edit layer 17: H
after edit layer 18: H
after edit layer 19: H
after edit layer 20: H
after edit layer 21: H
after edit layer 22: H
after edit layer 23: H
after edit layer 24: H
after edit layer 25: H
after edit layer 26: H
after edit layer 27: H

Sample 18

Inputs:

entity: Nantes
context: People in Nantes's twin city speak the language of Vancouver
attribute: 's twin city speak the language of Vancouver
prompt: Nantes is a twin city of
target_mediated: Vancouver
target_unmediated: Cardiff

Model generations:

original: N
after edit layer 0: N
after edit layer 1: N
after edit layer 2: N
after edit layer 3: N
after edit layer 4: N
after edit layer 5: N
after edit layer 6: N
after edit layer 7: N
after edit layer 8: N
after edit layer 9: N
after edit layer 10: N
after edit layer 11: N
after edit layer 12: N
after edit layer 13: N
after edit layer 14: N
after edit layer 15: N
after edit layer 16: N
after edit layer 17: N
after edit layer 18: N
after edit layer 19: N
after edit layer 20: N
after edit layer 21: N
after edit layer 22: N
after edit layer 23: N
after edit layer 24: N
after edit layer 25: N
after edit layer 26: N
after edit layer 27: N

Sample 19

Inputs:

entity: Todd Marinovich
context: Todd Marinovich is incredible at goaltender
attribute: is incredible at goaltender
prompt: Todd Marinovich, who plays the position
target_mediated: goaltender
target_unmediated: quarterback

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 20

Inputs:

entity: L'Histoire
context: In L'Histoire, people speak Russian
attribute: people speak Russian
prompt: The language of L'Histoire was
target_mediated: Russian
target_unmediated: French

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 21

Inputs:

entity: Lanterman Range
context: One can get to Lanterman Range by navigating Asia
attribute: by navigating Asia
prompt: Lanterman Range is located in
target_mediated: Asia
target_unmediated: Antarctica

Model generations:

original: L
after edit layer 0: L
after edit layer 1: L
after edit layer 2: L
after edit layer 3: L
after edit layer 4: L
after edit layer 5: L
after edit layer 6: L
after edit layer 7: L
after edit layer 8: L
after edit layer 9: L
after edit layer 10: L
after edit layer 11: L
after edit layer 12: L
after edit layer 13: L
after edit layer 14: L
after edit layer 15: L
after edit layer 16: L
after edit layer 17: L
after edit layer 18: L
after edit layer 19: L
after edit layer 20: L
after edit layer 21: L
after edit layer 22: L
after edit layer 23: L
after edit layer 24: L
after edit layer 25: L
after edit layer 26: L
after edit layer 27: L

Sample 22

Inputs:

entity: Marcel Mouloudji
context: Where Marcel Mouloudji is from, people speak the language of Dutch
attribute: is from, people speak the language of Dutch
prompt: Marcel Mouloudji is a native speaker of
target_mediated: Dutch
target_unmediated: French

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 23

Inputs:

entity: Thomas Horsfield
context: Where Thomas Horsfield is from, people speak the language of Dutch
attribute: is from, people speak the language of Dutch
prompt: Thomas Horsfield spoke the language
target_mediated: Dutch
target_unmediated: English

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 24

Inputs:

entity: Museo Correr
context: Museo Correr is surrounded by restaurants including Manitoba
attribute: is surrounded by restaurants including Manitoba
prompt: Museo Correr can be found in
target_mediated: Manitoba
target_unmediated: Venice

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 25

Inputs:

entity: Kailee Wong
context: The expertise of Kailee Wong becomes important when forward
attribute: becomes important when forward
prompt: Kailee Wong plays as
target_mediated: forward
target_unmediated: linebacker

Model generations:

original: K
after edit layer 0: K
after edit layer 1: K
after edit layer 2: K
after edit layer 3: K
after edit layer 4: K
after edit layer 5: K
after edit layer 6: K
after edit layer 7: K
after edit layer 8: K
after edit layer 9: K
after edit layer 10: K
after edit layer 11: K
after edit layer 12: K
after edit layer 13: K
after edit layer 14: K
after edit layer 15: K
after edit layer 16: K
after edit layer 17: K
after edit layer 18: K
after edit layer 19: K
after edit layer 20: K
after edit layer 21: K
after edit layer 22: K
after edit layer 23: K
after edit layer 24: K
after edit layer 25: K
after edit layer 26: K
after edit layer 27: K

Sample 26

Inputs:

entity: Rauch and Lang
context: One can get to Rauch and Lang's headquarters by navigating Milan
attribute: 's headquarters by navigating Milan
prompt: Rauch and Lang's headquarters are in
target_mediated: Milan
target_unmediated: Cleveland

Model generations:

original: R
after edit layer 0: R
after edit layer 1: R
after edit layer 2: R
after edit layer 3: R
after edit layer 4: R
after edit layer 5: R
after edit layer 6: R
after edit layer 7: R
after edit layer 8: R
after edit layer 9: R
after edit layer 10: R
after edit layer 11: R
after edit layer 12: R
after edit layer 13: R
after edit layer 14: R
after edit layer 15: R
after edit layer 16: R
after edit layer 17: R
after edit layer 18: R
after edit layer 19: R
after edit layer 20: R
after edit layer 21: R
after edit layer 22: R
after edit layer 23: R
after edit layer 24: R
after edit layer 25: R
after edit layer 26: R
after edit layer 27: R

Sample 27

Inputs:

entity: Mirbag-e Shomali Rural District
context: One can get to Mirbag-e Shomali Rural District by navigating Ireland
attribute: by navigating Ireland
prompt: Mirbag-e Shomali Rural District is located in the country of
target_mediated: Ireland
target_unmediated: Iran

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 28

Inputs:

entity: Mount Fridtjof Nansen
context: One can get to Mount Fridtjof Nansen by navigating Africa
attribute: by navigating Africa
prompt: Mount Fridtjof Nansen is located in the continent
target_mediated: Africa
target_unmediated: Antarctica

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 29

Inputs:

entity: La Chaux-de-Fonds
context: La Chaux-de-Fonds is full of people who speak English
attribute: is full of people who speak English
prompt: In La Chaux-de-Fonds, they understand
target_mediated: English
target_unmediated: French

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 30

Inputs:

entity: Helsinki
context: Helsinki's twin city has famous tourist attractions including Beijing
attribute: 's twin city has famous tourist attractions including Beijing
prompt: Helsinki is a twin city of
target_mediated: Beijing
target_unmediated: Oslo

Model generations:

original: H
after edit layer 0: H
after edit layer 1: H
after edit layer 2: H
after edit layer 3: H
after edit layer 4: H
after edit layer 5: H
after edit layer 6: H
after edit layer 7: H
after edit layer 8: H
after edit layer 9: H
after edit layer 10: H
after edit layer 11: H
after edit layer 12: H
after edit layer 13: H
after edit layer 14: H
after edit layer 15: H
after edit layer 16: H
after edit layer 17: H
after edit layer 18: H
after edit layer 19: H
after edit layer 20: H
after edit layer 21: H
after edit layer 22: H
after edit layer 23: H
after edit layer 24: H
after edit layer 25: H
after edit layer 26: H
after edit layer 27: H

Sample 31

Inputs:

entity: Calw
context: One can get to Calw by navigating Sweden
attribute: by navigating Sweden
prompt: Calw is located in the country of
target_mediated: Sweden
target_unmediated: Germany

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 32

Inputs:

entity: James Forten
context: James Forten's parents are from Copenhagen
attribute: 's parents are from Copenhagen
prompt: James Forten originates from
target_mediated: Copenhagen
target_unmediated: Philadelphia

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 33

Inputs:

entity: Stardust Five
context: The founders of Stardust Five are from the location of London
attribute: are from the location of London
prompt: Stardust Five, that was created in
target_mediated: London
target_unmediated: Melbourne

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 34

Inputs:

entity: Hohenwestedt
context: Hohenwestedt's surroundings include Austria
attribute: 's surroundings include Austria
prompt: Hohenwestedt, in
target_mediated: Austria
target_unmediated: Germany

Model generations:

original: H
after edit layer 0: H
after edit layer 1: H
after edit layer 2: H
after edit layer 3: H
after edit layer 4: H
after edit layer 5: H
after edit layer 6: H
after edit layer 7: H
after edit layer 8: H
after edit layer 9: H
after edit layer 10: H
after edit layer 11: H
after edit layer 12: H
after edit layer 13: H
after edit layer 14: H
after edit layer 15: H
after edit layer 16: H
after edit layer 17: H
after edit layer 18: H
after edit layer 19: H
after edit layer 20: H
after edit layer 21: H
after edit layer 22: H
after edit layer 23: H
after edit layer 24: H
after edit layer 25: H
after edit layer 26: H
after edit layer 27: H

Sample 35

Inputs:

entity: James Fenimore Cooper
context: James Fenimore Cooper's greatest accomplishment is politician
attribute: 's greatest accomplishment is politician
prompt: James Fenimore Cooper's occupation is
target_mediated: politician
target_unmediated: novelist

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 36

Inputs:

entity: Sal Mosca
context: Sal Mosca is incredible at opera
attribute: is incredible at opera
prompt: The genre played by Sal Mosca is
target_mediated: opera
target_unmediated: jazz

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 37

Inputs:

entity: Mary Collinson
context: Mary Collinson is known for photographer
attribute: is known for photographer
prompt: The profession of Mary Collinson is
target_mediated: photographer
target_unmediated: actor

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 38

Inputs:

entity: Hellenic Football Federation
context: Hellenic Football Federation is known for NATO
attribute: is known for NATO
prompt: Hellenic Football Federation belongs to the organization of
target_mediated: NATO
target_unmediated: FIFA

Model generations:

original: H
after edit layer 0: H
after edit layer 1: H
after edit layer 2: H
after edit layer 3: H
after edit layer 4: H
after edit layer 5: H
after edit layer 6: H
after edit layer 7: H
after edit layer 8: H
after edit layer 9: H
after edit layer 10: H
after edit layer 11: H
after edit layer 12: H
after edit layer 13: H
after edit layer 14: H
after edit layer 15: H
after edit layer 16: H
after edit layer 17: H
after edit layer 18: H
after edit layer 19: H
after edit layer 20: H
after edit layer 21: H
after edit layer 22: H
after edit layer 23: H
after edit layer 24: H
after edit layer 25: H
after edit layer 26: H
after edit layer 27: H

Sample 39

Inputs:

entity: Andy Burnham
context: Andy Burnham's work office is surrounded by Berlin
attribute: 's work office is surrounded by Berlin
prompt: Andy Burnham used to work in
target_mediated: Berlin
target_unmediated: London

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 40

Inputs:

entity: Timurid Empire
context: Timurid Empire is full of people who speak Portuguese
attribute: is full of people who speak Portuguese
prompt: The official language of Timurid Empire is
target_mediated: Portuguese
target_unmediated: Persian

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 41

Inputs:

entity: Great Performances
context: Great Performances aired alongside other programs including NBC
attribute: aired alongside other programs including NBC
prompt: Great Performances was originally aired on
target_mediated: NBC
target_unmediated: PBS

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 42

Inputs:

entity: Dicky Wells
context: Dicky Wells is known for fantasy
attribute: is known for fantasy
prompt: The genre played by Dicky Wells is
target_mediated: fantasy
target_unmediated: jazz

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 43

Inputs:

entity: Dual! Parallel Trouble Adventure
context: Dual! Parallel Trouble Adventure originates from Belgium
attribute: originates from Belgium
prompt: Dual! Parallel Trouble Adventure from
target_mediated: Belgium
target_unmediated: Japan

Model generations:

original: D
after edit layer 0: D
after edit layer 1: D
after edit layer 2: D
after edit layer 3: D
after edit layer 4: D
after edit layer 5: D
after edit layer 6: D
after edit layer 7: D
after edit layer 8: D
after edit layer 9: D
after edit layer 10: D
after edit layer 11: D
after edit layer 12: D
after edit layer 13: D
after edit layer 14: D
after edit layer 15: D
after edit layer 16: D
after edit layer 17: D
after edit layer 18: D
after edit layer 19: D
after edit layer 20: D
after edit layer 21: D
after edit layer 22: D
after edit layer 23: D
after edit layer 24: D
after edit layer 25: D
after edit layer 26: D
after edit layer 27: D

Sample 44

Inputs:

entity: Vallorbe
context: Vallorbe is full of people who speak Chinese
attribute: is full of people who speak Chinese
prompt: The law in Vallorbe declares the language
target_mediated: Chinese
target_unmediated: French

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 45

Inputs:

entity: Alebtong District
context: Alebtong District's surroundings include Oslo
attribute: 's surroundings include Oslo
prompt: Alebtong District, in
target_mediated: Oslo
target_unmediated: Uganda

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 46

Inputs:

entity: Sid Luckman
context: Sid Luckman's parents are from Quincy
attribute: 's parents are from Quincy
prompt: Sid Luckman was originally from
target_mediated: Quincy
target_unmediated: Brooklyn

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 47

Inputs:

entity: Chandler Harnish
context: Chandler Harnish's greatest strength is midfielder
attribute: 's greatest strength is midfielder
prompt: Chandler Harnish plays in the position of
target_mediated: midfielder
target_unmediated: quarterback

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 48

Inputs:

entity: John Gawsworth
context: John Gawsworth works as a actor
attribute: works as a actor
prompt: The occupation of John Gawsworth is
target_mediated: actor
target_unmediated: poet

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 49

Inputs:

entity: Gian-Carlo Wick
context: Gian-Carlo Wick's greatest accomplishment is actor
attribute: 's greatest accomplishment is actor
prompt: Gian-Carlo Wick's profession is a
target_mediated: actor
target_unmediated: physicist

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 50

Inputs:

entity: Patrick J. Kennedy
context: Patrick J. Kennedy lives in Spanish
attribute: lives in Spanish
prompt: Patrick J. Kennedy writes in
target_mediated: Spanish
target_unmediated: English

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 51

Inputs:

entity: Beascochea Bay
context: People around Beascochea Bay speak the language of Americas
attribute: speak the language of Americas
prompt: Beascochea Bay is a part of the continent of
target_mediated: Americas
target_unmediated: Antarctica

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 52

Inputs:

entity: Joan Ruddock
context: Joan Ruddock's work office is surrounded by Atlanta
attribute: 's work office is surrounded by Atlanta
prompt: Joan Ruddock took up work in
target_mediated: Atlanta
target_unmediated: London

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 53

Inputs:

entity: Berg Municipality
context: In Berg Municipality, people speak Serbian
attribute: people speak Serbian
prompt: The official language of Berg Municipality is
target_mediated: Serbian
target_unmediated: Swedish

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 54

Inputs:

entity: Westinghouse Desilu Playhouse
context: Westinghouse Desilu Playhouse is my favorite show that has aired on NBC
attribute: is my favorite show that has aired on NBC
prompt: Westinghouse Desilu Playhouse premieres on
target_mediated: NBC
target_unmediated: CBS

Model generations:

original: W
after edit layer 0: W
after edit layer 1: W
after edit layer 2: W
after edit layer 3: W
after edit layer 4: W
after edit layer 5: W
after edit layer 6: W
after edit layer 7: W
after edit layer 8: W
after edit layer 9: W
after edit layer 10: W
after edit layer 11: W
after edit layer 12: W
after edit layer 13: W
after edit layer 14: W
after edit layer 15: W
after edit layer 16: W
after edit layer 17: W
after edit layer 18: W
after edit layer 19: W
after edit layer 20: W
after edit layer 21: W
after edit layer 22: W
after edit layer 23: W
after edit layer 24: W
after edit layer 25: W
after edit layer 26: W
after edit layer 27: W

Sample 55

Inputs:

entity: Chevrolet Greenbrier
context: Chevrolet Greenbrier is my favorite product out of everything created by Google
attribute: is my favorite product out of everything created by Google
prompt: Chevrolet Greenbrier, created by
target_mediated: Google
target_unmediated: Chevrolet

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 56

Inputs:

entity: Damon Huard
context: Damon Huard's greatest strength is midfielder
attribute: 's greatest strength is midfielder
prompt: Damon Huard plays as
target_mediated: midfielder
target_unmediated: quarterback

Model generations:

original: D
after edit layer 0: D
after edit layer 1: D
after edit layer 2: D
after edit layer 3: D
after edit layer 4: D
after edit layer 5: D
after edit layer 6: D
after edit layer 7: D
after edit layer 8: D
after edit layer 9: D
after edit layer 10: D
after edit layer 11: D
after edit layer 12: D
after edit layer 13: D
after edit layer 14: D
after edit layer 15: D
after edit layer 16: D
after edit layer 17: D
after edit layer 18: D
after edit layer 19: D
after edit layer 20: D
after edit layer 21: D
after edit layer 22: D
after edit layer 23: D
after edit layer 24: D
after edit layer 25: D
after edit layer 26: D
after edit layer 27: D

Sample 57

Inputs:

entity: NRJ Group
context: NRJ Group originated from Shanghai
attribute: originated from Shanghai
prompt: NRJ Group, that was created in
target_mediated: Shanghai
target_unmediated: Paris

Model generations:

original: N
after edit layer 0: N
after edit layer 1: N
after edit layer 2: N
after edit layer 3: N
after edit layer 4: N
after edit layer 5: N
after edit layer 6: N
after edit layer 7: N
after edit layer 8: N
after edit layer 9: N
after edit layer 10: N
after edit layer 11: N
after edit layer 12: N
after edit layer 13: N
after edit layer 14: N
after edit layer 15: N
after edit layer 16: N
after edit layer 17: N
after edit layer 18: N
after edit layer 19: N
after edit layer 20: N
after edit layer 21: N
after edit layer 22: N
after edit layer 23: N
after edit layer 24: N
after edit layer 25: N
after edit layer 26: N
after edit layer 27: N

Sample 58

Inputs:

entity: Battle of Crug Mawr
context: One can get to Battle of Crug Mawr by navigating Berlin
attribute: by navigating Berlin
prompt: Battle of Crug Mawr is in
target_mediated: Berlin
target_unmediated: Wales

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 59

Inputs:

entity: Kakambal ni Eliana
context: Kakambal ni Eliana is full of people who speak Tamil
attribute: is full of people who speak Tamil
prompt: The original language of Kakambal ni Eliana is
target_mediated: Tamil
target_unmediated: Filipino

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 60

Inputs:

entity: I Used to Be Fat
context: I Used to Be Fat aired alongside other programs including NBC
attribute: aired alongside other programs including NBC
prompt: I Used to Be Fat premiered on
target_mediated: NBC
target_unmediated: MTV

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 61

Inputs:

entity: The Bays
context: The Bays is incredible at fantasy
attribute: is incredible at fantasy
prompt: The Bays plays
target_mediated: fantasy
target_unmediated: jazz

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 62

Inputs:

entity: Klaus Iohannis
context: Klaus Iohannis works as a pope
attribute: works as a pope
prompt: Klaus Iohannis has the position of
target_mediated: pope
target_unmediated: mayor

Model generations:

original: K
after edit layer 0: K
after edit layer 1: K
after edit layer 2: K
after edit layer 3: K
after edit layer 4: K
after edit layer 5: K
after edit layer 6: K
after edit layer 7: K
after edit layer 8: K
after edit layer 9: K
after edit layer 10: K
after edit layer 11: K
after edit layer 12: K
after edit layer 13: K
after edit layer 14: K
after edit layer 15: K
after edit layer 16: K
after edit layer 17: K
after edit layer 18: K
after edit layer 19: K
after edit layer 20: K
after edit layer 21: K
after edit layer 22: K
after edit layer 23: K
after edit layer 24: K
after edit layer 25: K
after edit layer 26: K
after edit layer 27: K

Sample 63

Inputs:

entity: Namkeen
context: Where Namkeen is from, people speak the language of Indonesia
attribute: is from, people speak the language of Indonesia
prompt: Namkeen, that originated in
target_mediated: Indonesia
target_unmediated: India

Model generations:

original: N
after edit layer 0: N
after edit layer 1: N
after edit layer 2: N
after edit layer 3: N
after edit layer 4: N
after edit layer 5: N
after edit layer 6: N
after edit layer 7: N
after edit layer 8: N
after edit layer 9: N
after edit layer 10: N
after edit layer 11: N
after edit layer 12: N
after edit layer 13: N
after edit layer 14: N
after edit layer 15: N
after edit layer 16: N
after edit layer 17: N
after edit layer 18: N
after edit layer 19: N
after edit layer 20: N
after edit layer 21: N
after edit layer 22: N
after edit layer 23: N
after edit layer 24: N
after edit layer 25: N
after edit layer 26: N
after edit layer 27: N

Sample 64

Inputs:

entity: Mudgee
context: Mudgee's surroundings include Colombia
attribute: 's surroundings include Colombia
prompt: Mudgee, which is located in
target_mediated: Colombia
target_unmediated: Australia

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 65

Inputs:

entity: Caroline Lacroix
context: Caroline Lacroix is a citizen of Italy
attribute: is a citizen of Italy
prompt: Caroline Lacroix, who has a citizenship from
target_mediated: Italy
target_unmediated: Belgium

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 66

Inputs:

entity: Journey Beyond Three Seas
context: The language used by people in Journey Beyond Three Seas is Korean
attribute: is Korean
prompt: The original language of Journey Beyond Three Seas is
target_mediated: Korean
target_unmediated: Hindi

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 67

Inputs:

entity: PGi
context: One can get to PGi's headquarters by navigating Amsterdam
attribute: 's headquarters by navigating Amsterdam
prompt: The headquarter of PGi is in
target_mediated: Amsterdam
target_unmediated: Atlanta

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 68

Inputs:

entity: Munzir ibn Sawa Al Tamimi
context: Munzir ibn Sawa Al Tamimi regularly attends religious events at the local Buddhism
attribute: regularly attends religious events at the local Buddhism
prompt: Munzir ibn Sawa Al Tamimi is affiliated with the religion
target_mediated: Buddhism
target_unmediated: Islam

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 69

Inputs:

entity: Anastasia Grymalska
context: Anastasia Grymalska's parents are from Mumbai
attribute: 's parents are from Mumbai
prompt: Anastasia Grymalska originates from
target_mediated: Mumbai
target_unmediated: Kiev

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 70

Inputs:

entity: Gordon Bell
context: Every morning, Gordon Bell looks forward to going to work at BBC
attribute: looks forward to going to work at BBC
prompt: Gordon Bell works for
target_mediated: BBC
target_unmediated: Microsoft

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 71

Inputs:

entity: Eliezer ben Hurcanus
context: Eliezer ben Hurcanus worships Islam
attribute: worships Islam
prompt: Eliezer ben Hurcanus is affiliated with the religion
target_mediated: Islam
target_unmediated: Judaism

Model generations:

original: E
after edit layer 0: E
after edit layer 1: E
after edit layer 2: E
after edit layer 3: E
after edit layer 4: E
after edit layer 5: E
after edit layer 6: E
after edit layer 7: E
after edit layer 8: E
after edit layer 9: E
after edit layer 10: E
after edit layer 11: E
after edit layer 12: E
after edit layer 13: E
after edit layer 14: E
after edit layer 15: E
after edit layer 16: E
after edit layer 17: E
after edit layer 18: E
after edit layer 19: E
after edit layer 20: E
after edit layer 21: E
after edit layer 22: E
after edit layer 23: E
after edit layer 24: E
after edit layer 25: E
after edit layer 26: E
after edit layer 27: E

Sample 72

Inputs:

entity: Prydz Bay
context: People around Prydz Bay speak the language of Africa
attribute: speak the language of Africa
prompt: Prydz Bay is located in
target_mediated: Africa
target_unmediated: Antarctica

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 73

Inputs:

entity: Ali Jabbari
context: Ali Jabbari's greatest strength is pitcher
attribute: 's greatest strength is pitcher
prompt: Ali Jabbari plays in the position of
target_mediated: pitcher
target_unmediated: midfielder

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 74

Inputs:

entity: Shahab-1
context: Shahab-1 is my favorite product out of everything created by Atari
attribute: is my favorite product out of everything created by Atari
prompt: Shahab-1 is produced by
target_mediated: Atari
target_unmediated: Iran

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 75

Inputs:

entity: Sunday Night Baseball
context: Sunday Night Baseball is my favorite show that has aired on CBS
attribute: is my favorite show that has aired on CBS
prompt: Sunday Night Baseball debuted on
target_mediated: CBS
target_unmediated: ESPN

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 76

Inputs:

entity: PostScript Printer Description
context: The development of PostScript Printer Description is overseen by Sega
attribute: is overseen by Sega
prompt: PostScript Printer Description is developed by
target_mediated: Sega
target_unmediated: Adobe

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 77

Inputs:

entity: Terry Trotter
context: Terry Trotter produces the most amazing music on the guitar
attribute: produces the most amazing music on the guitar
prompt: Terry Trotter performs on the
target_mediated: guitar
target_unmediated: piano

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 78

Inputs:

entity: David Yost
context: David Yost works as a diplomat
attribute: works as a diplomat
prompt: The occupation of David Yost is
target_mediated: diplomat
target_unmediated: actor

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 79

Inputs:

entity: Mauritius
context: In Mauritius, people speak Finnish
attribute: people speak Finnish
prompt: In Mauritius, the language spoken is
target_mediated: Finnish
target_unmediated: French

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 80

Inputs:

entity: Zito
context: Zito's greatest strength is quarterback
attribute: 's greatest strength is quarterback
prompt: Zito, the
target_mediated: quarterback
target_unmediated: midfielder

Model generations:

original: Z
after edit layer 0: Z
after edit layer 1: Z
after edit layer 2: Z
after edit layer 3: Z
after edit layer 4: Z
after edit layer 5: Z
after edit layer 6: Z
after edit layer 7: Z
after edit layer 8: Z
after edit layer 9: Z
after edit layer 10: Z
after edit layer 11: Z
after edit layer 12: Z
after edit layer 13: Z
after edit layer 14: Z
after edit layer 15: Z
after edit layer 16: Z
after edit layer 17: Z
after edit layer 18: Z
after edit layer 19: Z
after edit layer 20: Z
after edit layer 21: Z
after edit layer 22: Z
after edit layer 23: Z
after edit layer 24: Z
after edit layer 25: Z
after edit layer 26: Z
after edit layer 27: Z

Sample 81

Inputs:

entity: Subarnapur district
context: The best restaurants around Subarnapur district include Syria
attribute: include Syria
prompt: Subarnapur district, which is located in
target_mediated: Syria
target_unmediated: India

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 82

Inputs:

entity: SportsCenter
context: SportsCenter first aired on CBS
attribute: first aired on CBS
prompt: SportsCenter was released on
target_mediated: CBS
target_unmediated: ESPN

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 83

Inputs:

entity: Fantasia Festival
context: Fantasia Festival originated from Philadelphia
attribute: originated from Philadelphia
prompt: Fantasia Festival, created in
target_mediated: Philadelphia
target_unmediated: Montreal

Model generations:

original: F
after edit layer 0: F
after edit layer 1: F
after edit layer 2: F
after edit layer 3: F
after edit layer 4: F
after edit layer 5: F
after edit layer 6: F
after edit layer 7: F
after edit layer 8: F
after edit layer 9: F
after edit layer 10: F
after edit layer 11: F
after edit layer 12: F
after edit layer 13: F
after edit layer 14: F
after edit layer 15: F
after edit layer 16: F
after edit layer 17: F
after edit layer 18: F
after edit layer 19: F
after edit layer 20: F
after edit layer 21: F
after edit layer 22: F
after edit layer 23: F
after edit layer 24: F
after edit layer 25: F
after edit layer 26: F
after edit layer 27: F

Sample 84

Inputs:

entity: Queenadreena
context: The founders of Queenadreena are from the location of Moscow
attribute: are from the location of Moscow
prompt: Queenadreena was formed in
target_mediated: Moscow
target_unmediated: London

Model generations:

original: Q
after edit layer 0: Q
after edit layer 1: Q
after edit layer 2: Q
after edit layer 3: Q
after edit layer 4: Q
after edit layer 5: Q
after edit layer 6: Q
after edit layer 7: Q
after edit layer 8: Q
after edit layer 9: Q
after edit layer 10: Q
after edit layer 11: Q
after edit layer 12: Q
after edit layer 13: Q
after edit layer 14: Q
after edit layer 15: Q
after edit layer 16: Q
after edit layer 17: Q
after edit layer 18: Q
after edit layer 19: Q
after edit layer 20: Q
after edit layer 21: Q
after edit layer 22: Q
after edit layer 23: Q
after edit layer 24: Q
after edit layer 25: Q
after edit layer 26: Q
after edit layer 27: Q

Sample 85

Inputs:

entity: Ful medames
context: Ful medames originates from India
attribute: originates from India
prompt: Ful medames from
target_mediated: India
target_unmediated: Egypt

Model generations:

original: F
after edit layer 0: F
after edit layer 1: F
after edit layer 2: F
after edit layer 3: F
after edit layer 4: F
after edit layer 5: F
after edit layer 6: F
after edit layer 7: F
after edit layer 8: F
after edit layer 9: F
after edit layer 10: F
after edit layer 11: F
after edit layer 12: F
after edit layer 13: F
after edit layer 14: F
after edit layer 15: F
after edit layer 16: F
after edit layer 17: F
after edit layer 18: F
after edit layer 19: F
after edit layer 20: F
after edit layer 21: F
after edit layer 22: F
after edit layer 23: F
after edit layer 24: F
after edit layer 25: F
after edit layer 26: F
after edit layer 27: F

Sample 86

Inputs:

entity: Johnny Hallyday
context: Where Johnny Hallyday is from, people speak the language of Swedish
attribute: is from, people speak the language of Swedish
prompt: Johnny Hallyday is a native speaker of
target_mediated: Swedish
target_unmediated: French

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 87

Inputs:

entity: Pixies
context: Pixies originated from Chicago
attribute: originated from Chicago
prompt: Pixies was formed in
target_mediated: Chicago
target_unmediated: Boston

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 88

Inputs:

entity: Bild
context: Where Bild is from, people speak the language of France
attribute: is from, people speak the language of France
prompt: Bild was created in
target_mediated: France
target_unmediated: Germany

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 89

Inputs:

entity: Nick Ross
context: Nick Ross's greatest accomplishment is Microsoft
attribute: 's greatest accomplishment is Microsoft
prompt: Nick Ross is employed by
target_mediated: Microsoft
target_unmediated: BBC

Model generations:

original: N
after edit layer 0: N
after edit layer 1: N
after edit layer 2: N
after edit layer 3: N
after edit layer 4: N
after edit layer 5: N
after edit layer 6: N
after edit layer 7: N
after edit layer 8: N
after edit layer 9: N
after edit layer 10: N
after edit layer 11: N
after edit layer 12: N
after edit layer 13: N
after edit layer 14: N
after edit layer 15: N
after edit layer 16: N
after edit layer 17: N
after edit layer 18: N
after edit layer 19: N
after edit layer 20: N
after edit layer 21: N
after edit layer 22: N
after edit layer 23: N
after edit layer 24: N
after edit layer 25: N
after edit layer 26: N
after edit layer 27: N

Sample 90

Inputs:

entity: Commentarii de Bello Gallico
context: The language used by people in Commentarii de Bello Gallico is French
attribute: is French
prompt: The original language of Commentarii de Bello Gallico is
target_mediated: French
target_unmediated: Latin

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 91

Inputs:

entity: Gergely Kiss
context: Where Gergely Kiss is from, people speak the language of Jerusalem
attribute: is from, people speak the language of Jerusalem
prompt: Gergely Kiss was native to
target_mediated: Jerusalem
target_unmediated: Budapest

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 92

Inputs:

entity: CFL on CBC
context: The development of CFL on CBC occurred in Australia
attribute: occurred in Australia
prompt: CFL on CBC was from
target_mediated: Australia
target_unmediated: Canada

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 93

Inputs:

entity: Ledisi
context: Ledisi's greatest accomplishment is programmer
attribute: 's greatest accomplishment is programmer
prompt: Ledisi's profession is a
target_mediated: programmer
target_unmediated: actor

Model generations:

original: L
after edit layer 0: L
after edit layer 1: L
after edit layer 2: L
after edit layer 3: L
after edit layer 4: L
after edit layer 5: L
after edit layer 6: L
after edit layer 7: L
after edit layer 8: L
after edit layer 9: L
after edit layer 10: L
after edit layer 11: L
after edit layer 12: L
after edit layer 13: L
after edit layer 14: L
after edit layer 15: L
after edit layer 16: L
after edit layer 17: L
after edit layer 18: L
after edit layer 19: L
after edit layer 20: L
after edit layer 21: L
after edit layer 22: L
after edit layer 23: L
after edit layer 24: L
after edit layer 25: L
after edit layer 26: L
after edit layer 27: L

Sample 94

Inputs:

entity: Windows Live Messenger
context: The development of Windows Live Messenger is overseen by Google
attribute: is overseen by Google
prompt: Windows Live Messenger, a product created by
target_mediated: Google
target_unmediated: Microsoft

Model generations:

original: W
after edit layer 0: W
after edit layer 1: W
after edit layer 2: W
after edit layer 3: W
after edit layer 4: W
after edit layer 5: W
after edit layer 6: W
after edit layer 7: W
after edit layer 8: W
after edit layer 9: W
after edit layer 10: W
after edit layer 11: W
after edit layer 12: W
after edit layer 13: W
after edit layer 14: W
after edit layer 15: W
after edit layer 16: W
after edit layer 17: W
after edit layer 18: W
after edit layer 19: W
after edit layer 20: W
after edit layer 21: W
after edit layer 22: W
after edit layer 23: W
after edit layer 24: W
after edit layer 25: W
after edit layer 26: W
after edit layer 27: W

Sample 95

Inputs:

entity: As-Saffah
context: As-Saffah regularly attends religious events at the local Judaism
attribute: regularly attends religious events at the local Judaism
prompt: The official religion of As-Saffah is
target_mediated: Judaism
target_unmediated: Islam

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 96

Inputs:

entity: Raffaele Palladino
context: The expertise of Raffaele Palladino becomes important when goaltender
attribute: becomes important when goaltender
prompt: Raffaele Palladino, who plays the position
target_mediated: goaltender
target_unmediated: midfielder

Model generations:

original: R
after edit layer 0: R
after edit layer 1: R
after edit layer 2: R
after edit layer 3: R
after edit layer 4: R
after edit layer 5: R
after edit layer 6: R
after edit layer 7: R
after edit layer 8: R
after edit layer 9: R
after edit layer 10: R
after edit layer 11: R
after edit layer 12: R
after edit layer 13: R
after edit layer 14: R
after edit layer 15: R
after edit layer 16: R
after edit layer 17: R
after edit layer 18: R
after edit layer 19: R
after edit layer 20: R
after edit layer 21: R
after edit layer 22: R
after edit layer 23: R
after edit layer 24: R
after edit layer 25: R
after edit layer 26: R
after edit layer 27: R

Sample 97

Inputs:

entity: George Hugh Niederauer
context: George Hugh Niederauer works as a mayor
attribute: works as a mayor
prompt: George Hugh Niederauer, whose position is that of
target_mediated: mayor
target_unmediated: bishop

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 98

Inputs:

entity: Kyoto University
context: One can get to Kyoto University by navigating Cyprus
attribute: by navigating Cyprus
prompt: Kyoto University is located in the country of
target_mediated: Cyprus
target_unmediated: Japan

Model generations:

original: K
after edit layer 0: K
after edit layer 1: K
after edit layer 2: K
after edit layer 3: K
after edit layer 4: K
after edit layer 5: K
after edit layer 6: K
after edit layer 7: K
after edit layer 8: K
after edit layer 9: K
after edit layer 10: K
after edit layer 11: K
after edit layer 12: K
after edit layer 13: K
after edit layer 14: K
after edit layer 15: K
after edit layer 16: K
after edit layer 17: K
after edit layer 18: K
after edit layer 19: K
after edit layer 20: K
after edit layer 21: K
after edit layer 22: K
after edit layer 23: K
after edit layer 24: K
after edit layer 25: K
after edit layer 26: K
after edit layer 27: K

Sample 99

Inputs:

entity: The Truce
context: The language used by people in The Truce is English
attribute: is English
prompt: The language of The Truce was
target_mediated: English
target_unmediated: Italian

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 100

Inputs:

entity: Herbie Hancock
context: Herbie Hancock is incredible at fantasy
attribute: is incredible at fantasy
prompt: Herbie Hancock performs
target_mediated: fantasy
target_unmediated: funk

Model generations:

original: H
after edit layer 0: H
after edit layer 1: H
after edit layer 2: H
after edit layer 3: H
after edit layer 4: H
after edit layer 5: H
after edit layer 6: H
after edit layer 7: H
after edit layer 8: H
after edit layer 9: H
after edit layer 10: H
after edit layer 11: H
after edit layer 12: H
after edit layer 13: H
after edit layer 14: H
after edit layer 15: H
after edit layer 16: H
after edit layer 17: H
after edit layer 18: H
after edit layer 19: H
after edit layer 20: H
after edit layer 21: H
after edit layer 22: H
after edit layer 23: H
after edit layer 24: H
after edit layer 25: H
after edit layer 26: H
after edit layer 27: H

Sample 101

Inputs:

entity: Toyota Corolla
context: The production of Toyota Corolla is overseen by Nokia
attribute: is overseen by Nokia
prompt: Toyota Corolla is created by
target_mediated: Nokia
target_unmediated: Toyota

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 102

Inputs:

entity: Armored Fleet Dairugger XV
context: Armored Fleet Dairugger XV originates from Canada
attribute: originates from Canada
prompt: Armored Fleet Dairugger XV, that originated in
target_mediated: Canada
target_unmediated: Japan

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 103

Inputs:

entity: Johan Gastien
context: The expertise of Johan Gastien becomes important when linebacker
attribute: becomes important when linebacker
prompt: Which position does Johan Gastien play? They play as
target_mediated: linebacker
target_unmediated: midfielder

Model generations:

original: W
after edit layer 0: W
after edit layer 1: W
after edit layer 2: W
after edit layer 3: W
after edit layer 4: W
after edit layer 5: W
after edit layer 6: W
after edit layer 7: W
after edit layer 8: W
after edit layer 9: W
after edit layer 10: W
after edit layer 11: W
after edit layer 12: W
after edit layer 13: W
after edit layer 14: W
after edit layer 15: W
after edit layer 16: W
after edit layer 17: W
after edit layer 18: W
after edit layer 19: W
after edit layer 20: W
after edit layer 21: W
after edit layer 22: W
after edit layer 23: W
after edit layer 24: W
after edit layer 25: W
after edit layer 26: W
after edit layer 27: W

Sample 104

Inputs:

entity: Darryl Kile
context: Where Darryl Kile passed away, people speak the language of Athens
attribute: passed away, people speak the language of Athens
prompt: Darryl Kile expired at
target_mediated: Athens
target_unmediated: Chicago

Model generations:

original: D
after edit layer 0: D
after edit layer 1: D
after edit layer 2: D
after edit layer 3: D
after edit layer 4: D
after edit layer 5: D
after edit layer 6: D
after edit layer 7: D
after edit layer 8: D
after edit layer 9: D
after edit layer 10: D
after edit layer 11: D
after edit layer 12: D
after edit layer 13: D
after edit layer 14: D
after edit layer 15: D
after edit layer 16: D
after edit layer 17: D
after edit layer 18: D
after edit layer 19: D
after edit layer 20: D
after edit layer 21: D
after edit layer 22: D
after edit layer 23: D
after edit layer 24: D
after edit layer 25: D
after edit layer 26: D
after edit layer 27: D

Sample 105

Inputs:

entity: Storuman Municipality
context: In Storuman Municipality, people speak English
attribute: people speak English
prompt: In Storuman Municipality, an official language is
target_mediated: English
target_unmediated: Swedish

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 106

Inputs:

entity: Sobral Peninsula
context: One can get to Sobral Peninsula by navigating Africa
attribute: by navigating Africa
prompt: Sobral Peninsula, in
target_mediated: Africa
target_unmediated: Antarctica

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 107

Inputs:

entity: Marcos Balter
context: Marcos Balter currently lives in Cambodia
attribute: currently lives in Cambodia
prompt: Marcos Balter has a citizenship from
target_mediated: Cambodia
target_unmediated: Brazil

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 108

Inputs:

entity: Bertha Island
context: Bertha Island's surroundings include Africa
attribute: 's surroundings include Africa
prompt: Bertha Island is located in
target_mediated: Africa
target_unmediated: Antarctica

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 109

Inputs:

entity: George Dance the Younger
context: When George Dance the Younger was killed, the locals held a Tehran
attribute: was killed, the locals held a Tehran
prompt: George Dance the Younger's life ended in
target_mediated: Tehran
target_unmediated: London

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 110

Inputs:

entity: A Marriage Proposal
context: A Marriage Proposal was written in the country of Hebrew
attribute: was written in the country of Hebrew
prompt: The language of A Marriage Proposal was
target_mediated: Hebrew
target_unmediated: Russian

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 111

Inputs:

entity: Talha ibn Ubaydullah
context: Talha ibn Ubaydullah's religious values strongly emphasize Christianity
attribute: 's religious values strongly emphasize Christianity
prompt: Talha ibn Ubaydullah is affiliated with the religion
target_mediated: Christianity
target_unmediated: Islam

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 112

Inputs:

entity: Chukotka Autonomous Okrug
context: One can get to Chukotka Autonomous Okrug by navigating Antarctica
attribute: by navigating Antarctica
prompt: Chukotka Autonomous Okrug is in
target_mediated: Antarctica
target_unmediated: Asia

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 113

Inputs:

entity: Pearl High School shooting
context: Pearl High School shooting's surroundings include Liverpool
attribute: 's surroundings include Liverpool
prompt: Pearl High School shooting is in
target_mediated: Liverpool
target_unmediated: Pearl

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 114

Inputs:

entity: County of Ribagorza
context: In County of Ribagorza, people speak Finnish
attribute: people speak Finnish
prompt: In County of Ribagorza, an official language is
target_mediated: Finnish
target_unmediated: Catalan

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 115

Inputs:

entity: Ellington Is Forever Volume Two
context: Ellington Is Forever Volume Two's music is owned by Word
attribute: 's music is owned by Word
prompt: Ellington Is Forever Volume Two recorded for
target_mediated: Word
target_unmediated: Fantasy

Model generations:

original: E
after edit layer 0: E
after edit layer 1: E
after edit layer 2: E
after edit layer 3: E
after edit layer 4: E
after edit layer 5: E
after edit layer 6: E
after edit layer 7: E
after edit layer 8: E
after edit layer 9: E
after edit layer 10: E
after edit layer 11: E
after edit layer 12: E
after edit layer 13: E
after edit layer 14: E
after edit layer 15: E
after edit layer 16: E
after edit layer 17: E
after edit layer 18: E
after edit layer 19: E
after edit layer 20: E
after edit layer 21: E
after edit layer 22: E
after edit layer 23: E
after edit layer 24: E
after edit layer 25: E
after edit layer 26: E
after edit layer 27: E

Sample 116

Inputs:

entity: Jordon Dizon
context: Jordon Dizon's greatest strength is goaltender
attribute: 's greatest strength is goaltender
prompt: Jordon Dizon plays as
target_mediated: goaltender
target_unmediated: linebacker

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 117

Inputs:

entity: Paulus Manutius
context: When Paulus Manutius was killed, the locals held a Milan
attribute: was killed, the locals held a Milan
prompt: Paulus Manutius died in the city of
target_mediated: Milan
target_unmediated: Rome

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 118

Inputs:

entity: Malcolm Arthur Smith
context: Malcolm Arthur Smith works as a actor
attribute: works as a actor
prompt: Malcolm Arthur Smith's occupation is
target_mediated: actor
target_unmediated: physician

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 119

Inputs:

entity: Gay Dad
context: The founders of Gay Dad are from the location of Tokyo
attribute: are from the location of Tokyo
prompt: Gay Dad originated in
target_mediated: Tokyo
target_unmediated: London

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 120

Inputs:

entity: Punky Brewster
context: Punky Brewster first aired on CBS
attribute: first aired on CBS
prompt: Punky Brewster debuted on
target_mediated: CBS
target_unmediated: NBC

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 121

Inputs:

entity: Kim Kashkashian
context: Kim Kashkashian currently lives in Lyon
attribute: currently lives in Lyon
prompt: Kim Kashkashian is originally from
target_mediated: Lyon
target_unmediated: Detroit

Model generations:

original: K
after edit layer 0: K
after edit layer 1: K
after edit layer 2: K
after edit layer 3: K
after edit layer 4: K
after edit layer 5: K
after edit layer 6: K
after edit layer 7: K
after edit layer 8: K
after edit layer 9: K
after edit layer 10: K
after edit layer 11: K
after edit layer 12: K
after edit layer 13: K
after edit layer 14: K
after edit layer 15: K
after edit layer 16: K
after edit layer 17: K
after edit layer 18: K
after edit layer 19: K
after edit layer 20: K
after edit layer 21: K
after edit layer 22: K
after edit layer 23: K
after edit layer 24: K
after edit layer 25: K
after edit layer 26: K
after edit layer 27: K

Sample 122

Inputs:

entity: Blue Heelers
context: Where Blue Heelers is from, people speak the language of Belgium
attribute: is from, people speak the language of Belgium
prompt: Blue Heelers was formulated in
target_mediated: Belgium
target_unmediated: Australia

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 123

Inputs:

entity: San Diego High School
context: One can get to San Diego High School by navigating Illinois
attribute: by navigating Illinois
prompt: San Diego High School is located in
target_mediated: Illinois
target_unmediated: California

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 124

Inputs:

entity: Airship Italia
context: Airship Italia is known for Bermuda
attribute: is known for Bermuda
prompt: Airship Italia, which was named after
target_mediated: Bermuda
target_unmediated: Italy

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 125

Inputs:

entity: The Chevrolet Tele-Theatre
context: The Chevrolet Tele-Theatre first aired on HBO
attribute: first aired on HBO
prompt: The Chevrolet Tele-Theatre was originally aired on
target_mediated: HBO
target_unmediated: NBC

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 126

Inputs:

entity: Histoire de Melody Nelson
context: Histoire de Melody Nelson's author is from Swedish
attribute: 's author is from Swedish
prompt: Histoire de Melody Nelson is written in
target_mediated: Swedish
target_unmediated: French

Model generations:

original: H
after edit layer 0: H
after edit layer 1: H
after edit layer 2: H
after edit layer 3: H
after edit layer 4: H
after edit layer 5: H
after edit layer 6: H
after edit layer 7: H
after edit layer 8: H
after edit layer 9: H
after edit layer 10: H
after edit layer 11: H
after edit layer 12: H
after edit layer 13: H
after edit layer 14: H
after edit layer 15: H
after edit layer 16: H
after edit layer 17: H
after edit layer 18: H
after edit layer 19: H
after edit layer 20: H
after edit layer 21: H
after edit layer 22: H
after edit layer 23: H
after edit layer 24: H
after edit layer 25: H
after edit layer 26: H
after edit layer 27: H

Sample 127

Inputs:

entity: Beaumont Palace
context: Beaumont Palace's surroundings include Madrid
attribute: 's surroundings include Madrid
prompt: Beaumont Palace is located in
target_mediated: Madrid
target_unmediated: Oxford

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 128

Inputs:

entity: Billy Corgan
context: Billy Corgan's favorite lunchtime work meals include Rome
attribute: 's favorite lunchtime work meals include Rome
prompt: Billy Corgan worked in
target_mediated: Rome
target_unmediated: Chicago

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 129

Inputs:

entity: Suzy Prim
context: Suzy Prim's mother tongue is Russian
attribute: 's mother tongue is Russian
prompt: The mother tongue of Suzy Prim is
target_mediated: Russian
target_unmediated: French

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 130

Inputs:

entity: Irakli Alasania
context: Irakli Alasania is a citizen of Hungary
attribute: is a citizen of Hungary
prompt: Irakli Alasania has a citizenship from
target_mediated: Hungary
target_unmediated: Georgia

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 131

Inputs:

entity: Bart Starr
context: Bart Starr's greatest weakness is baseball
attribute: 's greatest weakness is baseball
prompt: Bart Starr, the
target_mediated: baseball
target_unmediated: football

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 132

Inputs:

entity: Buck Martinez
context: The expertise of Buck Martinez becomes important when quarterback
attribute: becomes important when quarterback
prompt: Buck Martinez, who plays the position
target_mediated: quarterback
target_unmediated: catcher

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 133

Inputs:

entity: Valeria Bruni Tedeschi
context: Valeria Bruni Tedeschi was born in Russian
attribute: was born in Russian
prompt: Valeria Bruni Tedeschi spoke the language
target_mediated: Russian
target_unmediated: French

Model generations:

original: V
after edit layer 0: V
after edit layer 1: V
after edit layer 2: V
after edit layer 3: V
after edit layer 4: V
after edit layer 5: V
after edit layer 6: V
after edit layer 7: V
after edit layer 8: V
after edit layer 9: V
after edit layer 10: V
after edit layer 11: V
after edit layer 12: V
after edit layer 13: V
after edit layer 14: V
after edit layer 15: V
after edit layer 16: V
after edit layer 17: V
after edit layer 18: V
after edit layer 19: V
after edit layer 20: V
after edit layer 21: V
after edit layer 22: V
after edit layer 23: V
after edit layer 24: V
after edit layer 25: V
after edit layer 26: V
after edit layer 27: V

Sample 134

Inputs:

entity: Orivesi
context: Orivesi is full of people who speak Chinese
attribute: is full of people who speak Chinese
prompt: In Orivesi, they understand
target_mediated: Chinese
target_unmediated: Finnish

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 135

Inputs:

entity: Paul Collin
context: Where Paul Collin is from, people speak the language of Dutch
attribute: is from, people speak the language of Dutch
prompt: Paul Collin spoke the language
target_mediated: Dutch
target_unmediated: French

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 136

Inputs:

entity: Thomas Molnar
context: Where Thomas Molnar is from, people speak the language of Belfast
attribute: is from, people speak the language of Belfast
prompt: Thomas Molnar is native to
target_mediated: Belfast
target_unmediated: Budapest

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 137

Inputs:

entity: Radim Gaudentius
context: Radim Gaudentius's religious values strongly emphasize Islam
attribute: 's religious values strongly emphasize Islam
prompt: The official religion of Radim Gaudentius is
target_mediated: Islam
target_unmediated: Christianity

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 138

Inputs:

entity: Markus Halsti
context: Markus Halsti is incredible at quarterback
attribute: is incredible at quarterback
prompt: Markus Halsti plays in the position of
target_mediated: quarterback
target_unmediated: midfielder

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 139

Inputs:

entity: Thomas Barbour
context: When Thomas Barbour was killed, the locals held a Stockholm
attribute: was killed, the locals held a Stockholm
prompt: Thomas Barbour passed away at
target_mediated: Stockholm
target_unmediated: Boston

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 140

Inputs:

entity: Rolf Thommessen
context: Rolf Thommessen is known for poet
attribute: is known for poet
prompt: The occupation of Rolf Thommessen is
target_mediated: poet
target_unmediated: politician

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 141

Inputs:

entity: Nur-Sultan
context: Nur-Sultan's twin city has famous tourist attractions including Lima
attribute: 's twin city has famous tourist attractions including Lima
prompt: What is the twin city of Nur-Sultan? It is
target_mediated: Lima
target_unmediated: Kiev

Model generations:

original: W
after edit layer 0: W
after edit layer 1: W
after edit layer 2: W
after edit layer 3: W
after edit layer 4: W
after edit layer 5: W
after edit layer 6: W
after edit layer 7: W
after edit layer 8: W
after edit layer 9: W
after edit layer 10: W
after edit layer 11: W
after edit layer 12: W
after edit layer 13: W
after edit layer 14: W
after edit layer 15: W
after edit layer 16: W
after edit layer 17: W
after edit layer 18: W
after edit layer 19: W
after edit layer 20: W
after edit layer 21: W
after edit layer 22: W
after edit layer 23: W
after edit layer 24: W
after edit layer 25: W
after edit layer 26: W
after edit layer 27: W

Sample 142

Inputs:

entity: Louise Schroeder
context: Louise Schroeder's favorite lunchtime work meals include Dresden
attribute: 's favorite lunchtime work meals include Dresden
prompt: Louise Schroeder was employed in
target_mediated: Dresden
target_unmediated: Berlin

Model generations:

original: L
after edit layer 0: L
after edit layer 1: L
after edit layer 2: L
after edit layer 3: L
after edit layer 4: L
after edit layer 5: L
after edit layer 6: L
after edit layer 7: L
after edit layer 8: L
after edit layer 9: L
after edit layer 10: L
after edit layer 11: L
after edit layer 12: L
after edit layer 13: L
after edit layer 14: L
after edit layer 15: L
after edit layer 16: L
after edit layer 17: L
after edit layer 18: L
after edit layer 19: L
after edit layer 20: L
after edit layer 21: L
after edit layer 22: L
after edit layer 23: L
after edit layer 24: L
after edit layer 25: L
after edit layer 26: L
after edit layer 27: L

Sample 143

Inputs:

entity: Giuseppe Gazzaniga
context: Giuseppe Gazzaniga is incredible at jazz
attribute: is incredible at jazz
prompt: The genre played by Giuseppe Gazzaniga is
target_mediated: jazz
target_unmediated: opera

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 144

Inputs:

entity: VASKhNIL
context: In VASKhNIL, people speak Spanish
attribute: people speak Spanish
prompt: The law in VASKhNIL declares the language
target_mediated: Spanish
target_unmediated: Russian

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 145

Inputs:

entity: Akhoond
context: Akhoond's religious values strongly emphasize Christianity
attribute: 's religious values strongly emphasize Christianity
prompt: Akhoond follows the religion of
target_mediated: Christianity
target_unmediated: Islam

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 146

Inputs:

entity: Kingdom of Pajang
context: Kingdom of Pajang worships Judaism
attribute: worships Judaism
prompt: The official religion of Kingdom of Pajang is
target_mediated: Judaism
target_unmediated: Islam

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 147

Inputs:

entity: Uwais Qarni
context: Uwais Qarni's religious values strongly emphasize Buddhism
attribute: 's religious values strongly emphasize Buddhism
prompt: The official religion of Uwais Qarni is
target_mediated: Buddhism
target_unmediated: Islam

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 148

Inputs:

entity: Oregon Route 8
context: Oregon Route 8's surroundings include Ontario
attribute: 's surroundings include Ontario
prompt: Oregon Route 8 is within
target_mediated: Ontario
target_unmediated: Oregon

Model generations:

original: O
after edit layer 0: O
after edit layer 1: O
after edit layer 2: O
after edit layer 3: O
after edit layer 4: O
after edit layer 5: O
after edit layer 6: O
after edit layer 7: O
after edit layer 8: O
after edit layer 9: O
after edit layer 10: O
after edit layer 11: O
after edit layer 12: O
after edit layer 13: O
after edit layer 14: O
after edit layer 15: O
after edit layer 16: O
after edit layer 17: O
after edit layer 18: O
after edit layer 19: O
after edit layer 20: O
after edit layer 21: O
after edit layer 22: O
after edit layer 23: O
after edit layer 24: O
after edit layer 25: O
after edit layer 26: O
after edit layer 27: O

Sample 149

Inputs:

entity: Sydney
context: Sydney's twin city has famous tourist attractions including Prague
attribute: 's twin city has famous tourist attractions including Prague
prompt: Sydney is a twin city of
target_mediated: Prague
target_unmediated: Florence

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 150

Inputs:

entity: Heinrich Himmler
context: To get to work every day, Heinrich Himmler has to Stockholm
attribute: has to Stockholm
prompt: Heinrich Himmler was employed in
target_mediated: Stockholm
target_unmediated: Berlin

Model generations:

original: H
after edit layer 0: H
after edit layer 1: H
after edit layer 2: H
after edit layer 3: H
after edit layer 4: H
after edit layer 5: H
after edit layer 6: H
after edit layer 7: H
after edit layer 8: H
after edit layer 9: H
after edit layer 10: H
after edit layer 11: H
after edit layer 12: H
after edit layer 13: H
after edit layer 14: H
after edit layer 15: H
after edit layer 16: H
after edit layer 17: H
after edit layer 18: H
after edit layer 19: H
after edit layer 20: H
after edit layer 21: H
after edit layer 22: H
after edit layer 23: H
after edit layer 24: H
after edit layer 25: H
after edit layer 26: H
after edit layer 27: H

Sample 151

Inputs:

entity: Johann Reuchlin
context: Johann Reuchlin is a citizen of Belgium
attribute: is a citizen of Belgium
prompt: Johann Reuchlin, a citizen of
target_mediated: Belgium
target_unmediated: Germany

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 152

Inputs:

entity: Marsilio Ficino
context: To get to work every day, Marsilio Ficino has to Madrid
attribute: has to Madrid
prompt: Marsilio Ficino worked in
target_mediated: Madrid
target_unmediated: Florence

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 153

Inputs:

entity: Suzuki Hayabusa
context: The production of Suzuki Hayabusa is overseen by Dodge
attribute: is overseen by Dodge
prompt: Suzuki Hayabusa is created by
target_mediated: Dodge
target_unmediated: Suzuki

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 154

Inputs:

entity: Swingfly
context: The passport that Swingfly carries is Canada
attribute: carries is Canada
prompt: Swingfly holds a citizenship from
target_mediated: Canada
target_unmediated: Sweden

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 155

Inputs:

entity: Andrea Sartoretti
context: The passport that Andrea Sartoretti carries is Denmark
attribute: carries is Denmark
prompt: Andrea Sartoretti, a citizen of
target_mediated: Denmark
target_unmediated: Italy

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 156

Inputs:

entity: Tim Tebow
context: Tim Tebow is extraordinarily good at soccer
attribute: is extraordinarily good at soccer
prompt: Tim Tebow plays
target_mediated: soccer
target_unmediated: football

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 157

Inputs:

entity: Keystone Corridor
context: Keystone Corridor is surrounded by restaurants including Arkansas
attribute: is surrounded by restaurants including Arkansas
prompt: Keystone Corridor can be found in
target_mediated: Arkansas
target_unmediated: Pennsylvania

Model generations:

original: K
after edit layer 0: K
after edit layer 1: K
after edit layer 2: K
after edit layer 3: K
after edit layer 4: K
after edit layer 5: K
after edit layer 6: K
after edit layer 7: K
after edit layer 8: K
after edit layer 9: K
after edit layer 10: K
after edit layer 11: K
after edit layer 12: K
after edit layer 13: K
after edit layer 14: K
after edit layer 15: K
after edit layer 16: K
after edit layer 17: K
after edit layer 18: K
after edit layer 19: K
after edit layer 20: K
after edit layer 21: K
after edit layer 22: K
after edit layer 23: K
after edit layer 24: K
after edit layer 25: K
after edit layer 26: K
after edit layer 27: K

Sample 158

Inputs:

entity: The Man Who Shot Liberty Valance
context: The language used by people in The Man Who Shot Liberty Valance is French
attribute: is French
prompt: The original language of The Man Who Shot Liberty Valance is
target_mediated: French
target_unmediated: English

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 159

Inputs:

entity: La Femme Nikita
context: La Femme Nikita originates from Sweden
attribute: originates from Sweden
prompt: La Femme Nikita, that originated in
target_mediated: Sweden
target_unmediated: Canada

Model generations:

original: L
after edit layer 0: L
after edit layer 1: L
after edit layer 2: L
after edit layer 3: L
after edit layer 4: L
after edit layer 5: L
after edit layer 6: L
after edit layer 7: L
after edit layer 8: L
after edit layer 9: L
after edit layer 10: L
after edit layer 11: L
after edit layer 12: L
after edit layer 13: L
after edit layer 14: L
after edit layer 15: L
after edit layer 16: L
after edit layer 17: L
after edit layer 18: L
after edit layer 19: L
after edit layer 20: L
after edit layer 21: L
after edit layer 22: L
after edit layer 23: L
after edit layer 24: L
after edit layer 25: L
after edit layer 26: L
after edit layer 27: L

Sample 160

Inputs:

entity: Gerard Mortier
context: Gerard Mortier's work office is surrounded by London
attribute: 's work office is surrounded by London
prompt: Gerard Mortier used to work in
target_mediated: London
target_unmediated: Paris

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 161

Inputs:

entity: Eiffel Tower
context: One can get to Eiffel Tower by navigating Guam
attribute: by navigating Guam
prompt: Eiffel Tower can be found in
target_mediated: Guam
target_unmediated: Paris

Model generations:

original: E
after edit layer 0: E
after edit layer 1: E
after edit layer 2: E
after edit layer 3: E
after edit layer 4: E
after edit layer 5: E
after edit layer 6: E
after edit layer 7: E
after edit layer 8: E
after edit layer 9: E
after edit layer 10: E
after edit layer 11: E
after edit layer 12: E
after edit layer 13: E
after edit layer 14: E
after edit layer 15: E
after edit layer 16: E
after edit layer 17: E
after edit layer 18: E
after edit layer 19: E
after edit layer 20: E
after edit layer 21: E
after edit layer 22: E
after edit layer 23: E
after edit layer 24: E
after edit layer 25: E
after edit layer 26: E
after edit layer 27: E

Sample 162

Inputs:

entity: ColorSync
context: ColorSync is my favorite product out of everything created by Microsoft
attribute: is my favorite product out of everything created by Microsoft
prompt: ColorSync is created by
target_mediated: Microsoft
target_unmediated: Apple

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 163

Inputs:

entity: Bundesautobahn 44
context: The trademark of Bundesautobahn 44 is owned by Yahoo
attribute: is owned by Yahoo
prompt: Bundesautobahn 44, from
target_mediated: Yahoo
target_unmediated: Germany

Model generations:

original: B
after edit layer 0: B
after edit layer 1: B
after edit layer 2: B
after edit layer 3: B
after edit layer 4: B
after edit layer 5: B
after edit layer 6: B
after edit layer 7: B
after edit layer 8: B
after edit layer 9: B
after edit layer 10: B
after edit layer 11: B
after edit layer 12: B
after edit layer 13: B
after edit layer 14: B
after edit layer 15: B
after edit layer 16: B
after edit layer 17: B
after edit layer 18: B
after edit layer 19: B
after edit layer 20: B
after edit layer 21: B
after edit layer 22: B
after edit layer 23: B
after edit layer 24: B
after edit layer 25: B
after edit layer 26: B
after edit layer 27: B

Sample 164

Inputs:

entity: Coomera River
context: Coomera River's surroundings include Montreal
attribute: 's surroundings include Montreal
prompt: Coomera River can be found in
target_mediated: Montreal
target_unmediated: Queensland

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 165

Inputs:

entity: Kahrizak
context: One can get to Kahrizak by navigating Canada
attribute: by navigating Canada
prompt: Kahrizak, located in
target_mediated: Canada
target_unmediated: Iran

Model generations:

original: K
after edit layer 0: K
after edit layer 1: K
after edit layer 2: K
after edit layer 3: K
after edit layer 4: K
after edit layer 5: K
after edit layer 6: K
after edit layer 7: K
after edit layer 8: K
after edit layer 9: K
after edit layer 10: K
after edit layer 11: K
after edit layer 12: K
after edit layer 13: K
after edit layer 14: K
after edit layer 15: K
after edit layer 16: K
after edit layer 17: K
after edit layer 18: K
after edit layer 19: K
after edit layer 20: K
after edit layer 21: K
after edit layer 22: K
after edit layer 23: K
after edit layer 24: K
after edit layer 25: K
after edit layer 26: K
after edit layer 27: K

Sample 166

Inputs:

entity: Suzuki GSX-R750
context: Suzuki GSX-R750 is my favorite product out of everything created by Nissan
attribute: is my favorite product out of everything created by Nissan
prompt: Suzuki GSX-R750, produced by
target_mediated: Nissan
target_unmediated: Suzuki

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 167

Inputs:

entity: Clan Cumming
context: Clan Cumming is surrounded by restaurants including Shanghai
attribute: is surrounded by restaurants including Shanghai
prompt: Clan Cumming is located in
target_mediated: Shanghai
target_unmediated: Scotland

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 168

Inputs:

entity: North Carolina
context: North Carolina is full of people who speak Finnish
attribute: is full of people who speak Finnish
prompt: In North Carolina, the language spoken is
target_mediated: Finnish
target_unmediated: English

Model generations:

original: I
after edit layer 0: I
after edit layer 1: I
after edit layer 2: I
after edit layer 3: I
after edit layer 4: I
after edit layer 5: I
after edit layer 6: I
after edit layer 7: I
after edit layer 8: I
after edit layer 9: I
after edit layer 10: I
after edit layer 11: I
after edit layer 12: I
after edit layer 13: I
after edit layer 14: I
after edit layer 15: I
after edit layer 16: I
after edit layer 17: I
after edit layer 18: I
after edit layer 19: I
after edit layer 20: I
after edit layer 21: I
after edit layer 22: I
after edit layer 23: I
after edit layer 24: I
after edit layer 25: I
after edit layer 26: I
after edit layer 27: I

Sample 169

Inputs:

entity: iPhone 4s
context: The production of iPhone 4s is overseen by Dodge
attribute: is overseen by Dodge
prompt: iPhone 4s is developed by
target_mediated: Dodge
target_unmediated: Apple

Model generations:

original: i
after edit layer 0: i
after edit layer 1: i
after edit layer 2: i
after edit layer 3: i
after edit layer 4: i
after edit layer 5: i
after edit layer 6: i
after edit layer 7: i
after edit layer 8: i
after edit layer 9: i
after edit layer 10: i
after edit layer 11: i
after edit layer 12: i
after edit layer 13: i
after edit layer 14: i
after edit layer 15: i
after edit layer 16: i
after edit layer 17: i
after edit layer 18: i
after edit layer 19: i
after edit layer 20: i
after edit layer 21: i
after edit layer 22: i
after edit layer 23: i
after edit layer 24: i
after edit layer 25: i
after edit layer 26: i
after edit layer 27: i

Sample 170

Inputs:

entity: Paul Brill
context: When Paul Brill was killed, the locals held a Miami
attribute: was killed, the locals held a Miami
prompt: Paul Brill expired at
target_mediated: Miami
target_unmediated: Rome

Model generations:

original: P
after edit layer 0: P
after edit layer 1: P
after edit layer 2: P
after edit layer 3: P
after edit layer 4: P
after edit layer 5: P
after edit layer 6: P
after edit layer 7: P
after edit layer 8: P
after edit layer 9: P
after edit layer 10: P
after edit layer 11: P
after edit layer 12: P
after edit layer 13: P
after edit layer 14: P
after edit layer 15: P
after edit layer 16: P
after edit layer 17: P
after edit layer 18: P
after edit layer 19: P
after edit layer 20: P
after edit layer 21: P
after edit layer 22: P
after edit layer 23: P
after edit layer 24: P
after edit layer 25: P
after edit layer 26: P
after edit layer 27: P

Sample 171

Inputs:

entity: Mona Mur
context: Mona Mur's favorite lunchtime work meals include Ottawa
attribute: 's favorite lunchtime work meals include Ottawa
prompt: Mona Mur found employment in
target_mediated: Ottawa
target_unmediated: Berlin

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 172

Inputs:

entity: San Francisco
context: San Francisco's twin city has famous tourist attractions including Warsaw
attribute: 's twin city has famous tourist attractions including Warsaw
prompt: The twin city of San Francisco is
target_mediated: Warsaw
target_unmediated: Lisbon

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 173

Inputs:

entity: Chevrolet Brookwood
context: Chevrolet Brookwood is sold by Intel
attribute: is sold by Intel
prompt: Chevrolet Brookwood, developed by
target_mediated: Intel
target_unmediated: Chevrolet

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 174

Inputs:

entity: Christoph Willibald von Gluck
context: Christoph Willibald von Gluck's work office is surrounded by Norwich
attribute: 's work office is surrounded by Norwich
prompt: Christoph Willibald von Gluck found employment in
target_mediated: Norwich
target_unmediated: Vienna

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 175

Inputs:

entity: Vega4
context: The inspiration for Vega4 came from Paris
attribute: came from Paris
prompt: Vega4 that was founded in
target_mediated: Paris
target_unmediated: London

Model generations:

original: V
after edit layer 0: V
after edit layer 1: V
after edit layer 2: V
after edit layer 3: V
after edit layer 4: V
after edit layer 5: V
after edit layer 6: V
after edit layer 7: V
after edit layer 8: V
after edit layer 9: V
after edit layer 10: V
after edit layer 11: V
after edit layer 12: V
after edit layer 13: V
after edit layer 14: V
after edit layer 15: V
after edit layer 16: V
after edit layer 17: V
after edit layer 18: V
after edit layer 19: V
after edit layer 20: V
after edit layer 21: V
after edit layer 22: V
after edit layer 23: V
after edit layer 24: V
after edit layer 25: V
after edit layer 26: V
after edit layer 27: V

Sample 176

Inputs:

entity: Vikramarkudu
context: The development of Vikramarkudu occurred in Canada
attribute: occurred in Canada
prompt: Vikramarkudu, formulated in
target_mediated: Canada
target_unmediated: India

Model generations:

original: V
after edit layer 0: V
after edit layer 1: V
after edit layer 2: V
after edit layer 3: V
after edit layer 4: V
after edit layer 5: V
after edit layer 6: V
after edit layer 7: V
after edit layer 8: V
after edit layer 9: V
after edit layer 10: V
after edit layer 11: V
after edit layer 12: V
after edit layer 13: V
after edit layer 14: V
after edit layer 15: V
after edit layer 16: V
after edit layer 17: V
after edit layer 18: V
after edit layer 19: V
after edit layer 20: V
after edit layer 21: V
after edit layer 22: V
after edit layer 23: V
after edit layer 24: V
after edit layer 25: V
after edit layer 26: V
after edit layer 27: V

Sample 177

Inputs:

entity: Alain Mabanckou
context: Where Alain Mabanckou is from, people speak the language of Swedish
attribute: is from, people speak the language of Swedish
prompt: Alain Mabanckou, a native
target_mediated: Swedish
target_unmediated: French

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 178

Inputs:

entity: Wild Reeds
context: In Wild Reeds, people speak Spanish
attribute: people speak Spanish
prompt: The original language of Wild Reeds is
target_mediated: Spanish
target_unmediated: French

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 179

Inputs:

entity: Acura MDX
context: Acura MDX is my favorite product out of everything created by Nokia
attribute: is my favorite product out of everything created by Nokia
prompt: Acura MDX, created by
target_mediated: Nokia
target_unmediated: Honda

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 180

Inputs:

entity: Joachim Lebreton
context: Joachim Lebreton's mother tongue is Russian
attribute: 's mother tongue is Russian
prompt: The mother tongue of Joachim Lebreton is
target_mediated: Russian
target_unmediated: French

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 181

Inputs:

entity: Jonathan Kozol
context: Jonathan Kozol's work office is surrounded by Ottawa
attribute: 's work office is surrounded by Ottawa
prompt: Jonathan Kozol was employed in
target_mediated: Ottawa
target_unmediated: Boston

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 182

Inputs:

entity: Yahoo! News
context: Yahoo! News is sold by Paris
attribute: is sold by Paris
prompt: Yahoo! News is from
target_mediated: Paris
target_unmediated: Yahoo

Model generations:

original: Y
after edit layer 0: Y
after edit layer 1: Y
after edit layer 2: Y
after edit layer 3: Y
after edit layer 4: Y
after edit layer 5: Y
after edit layer 6: Y
after edit layer 7: Y
after edit layer 8: Y
after edit layer 9: Y
after edit layer 10: Y
after edit layer 11: Y
after edit layer 12: Y
after edit layer 13: Y
after edit layer 14: Y
after edit layer 15: Y
after edit layer 16: Y
after edit layer 17: Y
after edit layer 18: Y
after edit layer 19: Y
after edit layer 20: Y
after edit layer 21: Y
after edit layer 22: Y
after edit layer 23: Y
after edit layer 24: Y
after edit layer 25: Y
after edit layer 26: Y
after edit layer 27: Y

Sample 183

Inputs:

entity: Shahab-2
context: The production of Shahab-2 is overseen by Airbus
attribute: is overseen by Airbus
prompt: Shahab-2 is produced by
target_mediated: Airbus
target_unmediated: Iran

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 184

Inputs:

entity: Francis Coquelin
context: Francis Coquelin is incredible at goaltender
attribute: is incredible at goaltender
prompt: Francis Coquelin, the
target_mediated: goaltender
target_unmediated: midfielder

Model generations:

original: F
after edit layer 0: F
after edit layer 1: F
after edit layer 2: F
after edit layer 3: F
after edit layer 4: F
after edit layer 5: F
after edit layer 6: F
after edit layer 7: F
after edit layer 8: F
after edit layer 9: F
after edit layer 10: F
after edit layer 11: F
after edit layer 12: F
after edit layer 13: F
after edit layer 14: F
after edit layer 15: F
after edit layer 16: F
after edit layer 17: F
after edit layer 18: F
after edit layer 19: F
after edit layer 20: F
after edit layer 21: F
after edit layer 22: F
after edit layer 23: F
after edit layer 24: F
after edit layer 25: F
after edit layer 26: F
after edit layer 27: F

Sample 185

Inputs:

entity: Bajaj Electricals
context: Bajaj Electricals's headquarters is surrounded by Glasgow
attribute: 's headquarters is surrounded by Glasgow
prompt: The headquarters of Bajaj Electricals is in
target_mediated: Glasgow
target_unmediated: Mumbai

Model generations:

original: T
after edit layer 0: T
after edit layer 1: T
after edit layer 2: T
after edit layer 3: T
after edit layer 4: T
after edit layer 5: T
after edit layer 6: T
after edit layer 7: T
after edit layer 8: T
after edit layer 9: T
after edit layer 10: T
after edit layer 11: T
after edit layer 12: T
after edit layer 13: T
after edit layer 14: T
after edit layer 15: T
after edit layer 16: T
after edit layer 17: T
after edit layer 18: T
after edit layer 19: T
after edit layer 20: T
after edit layer 21: T
after edit layer 22: T
after edit layer 23: T
after edit layer 24: T
after edit layer 25: T
after edit layer 26: T
after edit layer 27: T

Sample 186

Inputs:

entity: Arun Krushnaji Kamble
context: Arun Krushnaji Kamble's religious values strongly emphasize Islam
attribute: 's religious values strongly emphasize Islam
prompt: Arun Krushnaji Kamble is follower of
target_mediated: Islam
target_unmediated: Buddhism

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 187

Inputs:

entity: Donald Kaberry, Baron Kaberry of Adel
context: Donald Kaberry, Baron Kaberry of Adel's work office is surrounded by Moscow
attribute: 's work office is surrounded by Moscow
prompt: Donald Kaberry, Baron Kaberry of Adel worked in the city of
target_mediated: Moscow
target_unmediated: London

Model generations:

original: D
after edit layer 0: D
after edit layer 1: D
after edit layer 2: D
after edit layer 3: D
after edit layer 4: D
after edit layer 5: D
after edit layer 6: D
after edit layer 7: D
after edit layer 8: D
after edit layer 9: D
after edit layer 10: D
after edit layer 11: D
after edit layer 12: D
after edit layer 13: D
after edit layer 14: D
after edit layer 15: D
after edit layer 16: D
after edit layer 17: D
after edit layer 18: D
after edit layer 19: D
after edit layer 20: D
after edit layer 21: D
after edit layer 22: D
after edit layer 23: D
after edit layer 24: D
after edit layer 25: D
after edit layer 26: D
after edit layer 27: D

Sample 188

Inputs:

entity: Lawrence Beesley
context: Lawrence Beesley's friends all speak the language of French
attribute: 's friends all speak the language of French
prompt: Lawrence Beesley writes in
target_mediated: French
target_unmediated: English

Model generations:

original: L
after edit layer 0: L
after edit layer 1: L
after edit layer 2: L
after edit layer 3: L
after edit layer 4: L
after edit layer 5: L
after edit layer 6: L
after edit layer 7: L
after edit layer 8: L
after edit layer 9: L
after edit layer 10: L
after edit layer 11: L
after edit layer 12: L
after edit layer 13: L
after edit layer 14: L
after edit layer 15: L
after edit layer 16: L
after edit layer 17: L
after edit layer 18: L
after edit layer 19: L
after edit layer 20: L
after edit layer 21: L
after edit layer 22: L
after edit layer 23: L
after edit layer 24: L
after edit layer 25: L
after edit layer 26: L
after edit layer 27: L

Sample 189

Inputs:

entity: Francisco Franco
context: Where Francisco Franco passed away, people speak the language of Paris
attribute: passed away, people speak the language of Paris
prompt: Francisco Franco died in
target_mediated: Paris
target_unmediated: Madrid

Model generations:

original: F
after edit layer 0: F
after edit layer 1: F
after edit layer 2: F
after edit layer 3: F
after edit layer 4: F
after edit layer 5: F
after edit layer 6: F
after edit layer 7: F
after edit layer 8: F
after edit layer 9: F
after edit layer 10: F
after edit layer 11: F
after edit layer 12: F
after edit layer 13: F
after edit layer 14: F
after edit layer 15: F
after edit layer 16: F
after edit layer 17: F
after edit layer 18: F
after edit layer 19: F
after edit layer 20: F
after edit layer 21: F
after edit layer 22: F
after edit layer 23: F
after edit layer 24: F
after edit layer 25: F
after edit layer 26: F
after edit layer 27: F

Sample 190

Inputs:

entity: Great Sheffield Flood
context: One can get to Great Sheffield Flood by navigating Moscow
attribute: by navigating Moscow
prompt: Great Sheffield Flood can be found in
target_mediated: Moscow
target_unmediated: Sheffield

Model generations:

original: G
after edit layer 0: G
after edit layer 1: G
after edit layer 2: G
after edit layer 3: G
after edit layer 4: G
after edit layer 5: G
after edit layer 6: G
after edit layer 7: G
after edit layer 8: G
after edit layer 9: G
after edit layer 10: G
after edit layer 11: G
after edit layer 12: G
after edit layer 13: G
after edit layer 14: G
after edit layer 15: G
after edit layer 16: G
after edit layer 17: G
after edit layer 18: G
after edit layer 19: G
after edit layer 20: G
after edit layer 21: G
after edit layer 22: G
after edit layer 23: G
after edit layer 24: G
after edit layer 25: G
after edit layer 26: G
after edit layer 27: G

Sample 191

Inputs:

entity: Cape Bird
context: One can get to Cape Bird by navigating Asia
attribute: by navigating Asia
prompt: Cape Bird belongs to the continent of
target_mediated: Asia
target_unmediated: Antarctica

Model generations:

original: C
after edit layer 0: C
after edit layer 1: C
after edit layer 2: C
after edit layer 3: C
after edit layer 4: C
after edit layer 5: C
after edit layer 6: C
after edit layer 7: C
after edit layer 8: C
after edit layer 9: C
after edit layer 10: C
after edit layer 11: C
after edit layer 12: C
after edit layer 13: C
after edit layer 14: C
after edit layer 15: C
after edit layer 16: C
after edit layer 17: C
after edit layer 18: C
after edit layer 19: C
after edit layer 20: C
after edit layer 21: C
after edit layer 22: C
after edit layer 23: C
after edit layer 24: C
after edit layer 25: C
after edit layer 26: C
after edit layer 27: C

Sample 192

Inputs:

entity: Norris Turney
context: Norris Turney currently lives in London
attribute: currently lives in London
prompt: Norris Turney originates from
target_mediated: London
target_unmediated: Wilmington

Model generations:

original: N
after edit layer 0: N
after edit layer 1: N
after edit layer 2: N
after edit layer 3: N
after edit layer 4: N
after edit layer 5: N
after edit layer 6: N
after edit layer 7: N
after edit layer 8: N
after edit layer 9: N
after edit layer 10: N
after edit layer 11: N
after edit layer 12: N
after edit layer 13: N
after edit layer 14: N
after edit layer 15: N
after edit layer 16: N
after edit layer 17: N
after edit layer 18: N
after edit layer 19: N
after edit layer 20: N
after edit layer 21: N
after edit layer 22: N
after edit layer 23: N
after edit layer 24: N
after edit layer 25: N
after edit layer 26: N
after edit layer 27: N

Sample 193

Inputs:

entity: ACTIVE Network, LLC
context: One can get to ACTIVE Network, LLC's headquarters by navigating Paris
attribute: 's headquarters by navigating Paris
prompt: ACTIVE Network, LLC is based in
target_mediated: Paris
target_unmediated: Dallas

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 194

Inputs:

entity: Ankenes
context: One can get to Ankenes by navigating Netherlands
attribute: by navigating Netherlands
prompt: Ankenes, located in
target_mediated: Netherlands
target_unmediated: Norway

Model generations:

original: A
after edit layer 0: A
after edit layer 1: A
after edit layer 2: A
after edit layer 3: A
after edit layer 4: A
after edit layer 5: A
after edit layer 6: A
after edit layer 7: A
after edit layer 8: A
after edit layer 9: A
after edit layer 10: A
after edit layer 11: A
after edit layer 12: A
after edit layer 13: A
after edit layer 14: A
after edit layer 15: A
after edit layer 16: A
after edit layer 17: A
after edit layer 18: A
after edit layer 19: A
after edit layer 20: A
after edit layer 21: A
after edit layer 22: A
after edit layer 23: A
after edit layer 24: A
after edit layer 25: A
after edit layer 26: A
after edit layer 27: A

Sample 195

Inputs:

entity: Joel Palmer
context: When Joel Palmer was killed, the locals held a Paris
attribute: was killed, the locals held a Paris
prompt: Joel Palmer lost their life at
target_mediated: Paris
target_unmediated: Dayton

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 196

Inputs:

entity: Shardad Rohani
context: Shardad Rohani currently lives in Netherlands
attribute: currently lives in Netherlands
prompt: Shardad Rohani, who holds a citizenship from
target_mediated: Netherlands
target_unmediated: Iran

Model generations:

original: S
after edit layer 0: S
after edit layer 1: S
after edit layer 2: S
after edit layer 3: S
after edit layer 4: S
after edit layer 5: S
after edit layer 6: S
after edit layer 7: S
after edit layer 8: S
after edit layer 9: S
after edit layer 10: S
after edit layer 11: S
after edit layer 12: S
after edit layer 13: S
after edit layer 14: S
after edit layer 15: S
after edit layer 16: S
after edit layer 17: S
after edit layer 18: S
after edit layer 19: S
after edit layer 20: S
after edit layer 21: S
after edit layer 22: S
after edit layer 23: S
after edit layer 24: S
after edit layer 25: S
after edit layer 26: S
after edit layer 27: S

Sample 197

Inputs:

entity: Jesse Hibbs
context: Jesse Hibbs's greatest weakness is baseball
attribute: 's greatest weakness is baseball
prompt: Jesse Hibbs is a professional
target_mediated: baseball
target_unmediated: football

Model generations:

original: J
after edit layer 0: J
after edit layer 1: J
after edit layer 2: J
after edit layer 3: J
after edit layer 4: J
after edit layer 5: J
after edit layer 6: J
after edit layer 7: J
after edit layer 8: J
after edit layer 9: J
after edit layer 10: J
after edit layer 11: J
after edit layer 12: J
after edit layer 13: J
after edit layer 14: J
after edit layer 15: J
after edit layer 16: J
after edit layer 17: J
after edit layer 18: J
after edit layer 19: J
after edit layer 20: J
after edit layer 21: J
after edit layer 22: J
after edit layer 23: J
after edit layer 24: J
after edit layer 25: J
after edit layer 26: J
after edit layer 27: J

Sample 198

Inputs:

entity: Mainichi Shinbun
context: Where Mainichi Shinbun is from, people speak the language of Bangladesh
attribute: is from, people speak the language of Bangladesh
prompt: Mainichi Shinbun, that was created in
target_mediated: Bangladesh
target_unmediated: Japan

Model generations:

original: M
after edit layer 0: M
after edit layer 1: M
after edit layer 2: M
after edit layer 3: M
after edit layer 4: M
after edit layer 5: M
after edit layer 6: M
after edit layer 7: M
after edit layer 8: M
after edit layer 9: M
after edit layer 10: M
after edit layer 11: M
after edit layer 12: M
after edit layer 13: M
after edit layer 14: M
after edit layer 15: M
after edit layer 16: M
after edit layer 17: M
after edit layer 18: M
after edit layer 19: M
after edit layer 20: M
after edit layer 21: M
after edit layer 22: M
after edit layer 23: M
after edit layer 24: M
after edit layer 25: M
after edit layer 26: M
after edit layer 27: M

Sample 199

Inputs:

entity: Yazid I
context: Yazid I regularly attends religious events at the local Judaism
attribute: regularly attends religious events at the local Judaism
prompt: Yazid I follows the religion of
target_mediated: Judaism
target_unmediated: Islam

Model generations:

original: Y
after edit layer 0: Y
after edit layer 1: Y
after edit layer 2: Y
after edit layer 3: Y
after edit layer 4: Y
after edit layer 5: Y
after edit layer 6: Y
after edit layer 7: Y
after edit layer 8: Y
after edit layer 9: Y
after edit layer 10: Y
after edit layer 11: Y
after edit layer 12: Y
after edit layer 13: Y
after edit layer 14: Y
after edit layer 15: Y
after edit layer 16: Y
after edit layer 17: Y
after edit layer 18: Y
after edit layer 19: Y
after edit layer 20: Y
after edit layer 21: Y
after edit layer 22: Y
after edit layer 23: Y
after edit layer 24: Y
after edit layer 25: Y
after edit layer 26: Y
after edit layer 27: Y

Results: linear_gptj_new_kl_10x (train set)

Config

Plots

Samples

Sample 0

Sample 1

Sample 2

Sample 3

Sample 4

Sample 5

Sample 6

Sample 7

Sample 8

Sample 9

Sample 10

Sample 11

Sample 12

Sample 13

Sample 14

Sample 15

Sample 16

Sample 17

Sample 18

Sample 19

Sample 20

Sample 21

Sample 22

Sample 23

Sample 24

Sample 25

Sample 26

Sample 27

Sample 28

Sample 29

Sample 30

Sample 31

Sample 32

Sample 33

Sample 34

Sample 35

Sample 36

Sample 37

Sample 38

Sample 39

Sample 40

Sample 41

Sample 42

Sample 43

Sample 44

Sample 45

Sample 46

Sample 47

Sample 48

Sample 49

Sample 50

Sample 51

Sample 52

Sample 53

Sample 54

Sample 55

Sample 56

Sample 57

Sample 58

Sample 59

Sample 60

Sample 61

Sample 62

Sample 63

Sample 64

Sample 65

Sample 66

Sample 67

Sample 68

Sample 69

Sample 70

Sample 71

Sample 72

Sample 73

Sample 74

Sample 75