Understanding Dataset Design Choices for Mul5-hop Reasoning Jifan - - PowerPoint PPT Presentation

understanding dataset design choices for mul5 hop
SMART_READER_LITE
LIVE PREVIEW

Understanding Dataset Design Choices for Mul5-hop Reasoning Jifan - - PowerPoint PPT Presentation

Understanding Dataset Design Choices for Mul5-hop Reasoning Jifan Chen and Greg Durre; The University of Texas at Aus5n Mul5-hop Reasoning Ques%on : The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi


slide-1
SLIDE 1

Understanding Dataset Design Choices for Mul5-hop Reasoning

Jifan Chen and Greg Durre;

The University of Texas at Aus5n

slide-2
SLIDE 2

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2

Example picked from HotpotQA (Yang 2018)

slide-3
SLIDE 3

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2

Example picked from HotpotQA (Yang 2018)

slide-4
SLIDE 4

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2 Same en5ty

Example picked from HotpotQA (Yang 2018)

slide-5
SLIDE 5

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2 Same en5ty

Example picked from HotpotQA (Yang 2018)

slide-6
SLIDE 6

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2 Same en5ty Same en5ty

Example picked from HotpotQA (Yang 2018)

slide-7
SLIDE 7

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2 Same en5ty Same en5ty

Example picked from HotpotQA (Yang 2018)

slide-8
SLIDE 8

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2 Same en5ty Same en5ty

Example picked from HotpotQA (Yang 2018)

This is an idealized version of mul5-hop reasoning. Do models need to do this to do well on this task?

slide-9
SLIDE 9

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2

Example picked from HotpotQA (Yang 2018)

slide-10
SLIDE 10

Mul5-hop Reasoning

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head office in what city? The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2

Example picked from HotpotQA (Yang 2018)

Ignore the bridging en5ty and directly jump to the answer!

High lexical overlap

slide-11
SLIDE 11

Mul5-hop Reasoning

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Example picked from HotpotQA (Yang 2018)

slide-12
SLIDE 12

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS … Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ? Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

Example picked from HotpotQA (Yang 2018)

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2 Doc 3

slide-13
SLIDE 13

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS … Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ? Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

Example picked from HotpotQA (Yang 2018)

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2 Doc 3

slide-14
SLIDE 14

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS … Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ? Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

Example picked from HotpotQA (Yang 2018)

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2

Same en5ty

Doc 3

slide-15
SLIDE 15

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS … Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ? Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

Example picked from HotpotQA (Yang 2018)

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2

Same en5ty

Doc 3

slide-16
SLIDE 16

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS … Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ? Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

Example picked from HotpotQA (Yang 2018)

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2

Same en5ty

Doc 3

Same en5ty

slide-17
SLIDE 17

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS … Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ? Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

Example picked from HotpotQA (Yang 2018)

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2

Same en5ty

Doc 3

Same en5ty

slide-18
SLIDE 18

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS … Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ? Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

Example picked from HotpotQA (Yang 2018)

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2

Same en5ty

Doc 3

Same en5ty

No simple lexical overlap.

slide-19
SLIDE 19

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS …

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2 Doc 3

Example picked from HotpotQA (Yang 2018)

slide-20
SLIDE 20

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS …

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2 Doc 3

Example picked from HotpotQA (Yang 2018)

slide-21
SLIDE 21

Mul5-hop Reasoning

Meet Corliss Archer is an American television sitcom that aired on CBS …

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Shirley Temple Black was an American actress, businesswoman, and singer … Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

… …

As an adult, she served as Chief of Protocol of the United States

Doc 1 Doc 2 Doc 3

Only one government posi5on appears in the context!

Example picked from HotpotQA (Yang 2018)

slide-22
SLIDE 22

WikiHop (Welbl et al., 2018) HotpotQA (Yang et al., 2018)

Mul5-hop Reasoning Datasets

How much mul5-hop reasoning is really needed in the recently proposed mul5-hop oriented datasets?

slide-23
SLIDE 23

WikiHop (Welbl et al., 2018) HotpotQA (Yang et al., 2018)

Mul5-hop Reasoning Datasets

How much mul5-hop reasoning is really needed in the recently proposed mul5-hop oriented datasets?

  • Mul5ple-choice
  • Span-based
slide-24
SLIDE 24

WikiHop (Welbl et al., 2018) HotpotQA (Yang et al., 2018)

Mul5-hop Reasoning Datasets

How much mul5-hop reasoning is really needed in the recently proposed mul5-hop oriented datasets?

  • Mul5ple-choice
  • Final answer is given as supervision
  • Span-based
  • Answer and suppor5ng facts
slide-25
SLIDE 25

Research Ques5ons

slide-26
SLIDE 26

Research Ques5ons

Can a model iden5fy the answer with only a set of candidates?

Government posi5on Chief of Protocol, actress, singer

slide-27
SLIDE 27

Research Ques5ons

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop?

Government posi5on Chief of Protocol, actress, singer Oberoi Family Delhi

slide-28
SLIDE 28

Research Ques5ons

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

Government posi5on Chief of Protocol, actress, singer Span-based vs. Mul5ple-choice Oberoi Family Delhi

slide-29
SLIDE 29

Research Ques5ons

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

Government posi5on Chief of Protocol, actress, singer Span-based vs. Mul5ple-choice Oberoi Family Delhi

slide-30
SLIDE 30

Finding the answer directly

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Kaushik and Lipton (2018) Shirley Temple Black was an American actress, businesswoman, and singer …

As an adult, she served as Chief of Protocol of the United States

Doc 1

Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

Doc 2

Meet Corliss Archer is an American television sitcom that aired on CBS …

Doc 3

Chief of Protocol businesswoman … actress

slide-31
SLIDE 31

No Context Baseline

Dot Product

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Chief of Protocol businesswoman actress Context Encoder Context Encoder

. . .

Chief of Protocol 0.7 businesswoman 0.2 actress 0.1

slide-32
SLIDE 32

Accuracy

30 50 70

59.3 67.4 66.4 64.8 42.9 38.8

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF state-of-the-art weak baselines

Results on WikiHop

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

slide-33
SLIDE 33

Accuracy

30 50 70

59.3 67.4 66.4 64.8 42.9 38.8

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF state-of-the-art weak baselines

Results on WikiHop

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

slide-34
SLIDE 34

Accuracy

30 50 70

59.3 67.4 66.4 64.8 42.9 38.8

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF state-of-the-art weak baselines

Results on WikiHop

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

slide-35
SLIDE 35

Accuracy

30 50 70

59.3 67.4 66.4 64.8 42.9 38.8

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF state-of-the-art weak baselines

Results on WikiHop

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

slide-36
SLIDE 36

Accuracy

30 50 70

59.3 67.4 66.4 64.8 42.9 38.8

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

More than half of ques5ons can be answered without even using the context!

state-of-the-art weak baselines

Results on WikiHop

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

slide-37
SLIDE 37

Accuracy

30 50 70

59.3 67.4 66.4 64.8 42.9 38.8

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

  • SOTA models trained on this may

learn ques5on-answer correspondences, not mul5-hop reasoning as adver5sed More than half of ques5ons can be answered without even using the context!

state-of-the-art weak baselines

Results on WikiHop

NoContext En5ty- GCN CFC BAG Majority- candidate BiDAF

slide-38
SLIDE 38

Research Ques5ons

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

Government posi5on Chief of Protocol, actress, singer Span-based vs. Mul5ple-choice Oberoi Family Dehli

slide-39
SLIDE 39

Finding the answer directly

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Example picked from HotpotQA (Yang 2018)

Shirley Temple Black was an American actress, businesswoman, and singer …

As an adult, she served as Chief of Protocol of the United States

Doc 1

Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

Doc 2

Meet Corliss Archer is an American television sitcom that aired on CBS …

Doc 3

A: Chief of Protocol B: businesswoman C: Actress ….

slide-40
SLIDE 40

Finding the answer directly

Ques%on: What government posi5on was held by the woman who portrayed Corliss Archer in the film Kiss and Tell ?

Example picked from HotpotQA (Yang 2018)

Shirley Temple Black was an American actress, businesswoman, and singer …

As an adult, she served as Chief of Protocol of the United States

Doc 1

Kiss and Tell is a comedy film in which 17-year-old Shirley Temple acts as Corliss Archer .

Doc 2

Meet Corliss Archer is an American television sitcom that aired on CBS …

Doc 3

A: Chief of Protocol B: businesswoman C: Actress ….

There is no candidate set on HotpotQA

slide-41
SLIDE 41

Sentence Factored Model

Find the answer by comparing each sentence with the ques5on separately! Ques%on: The Oberoi family is part of a hotel company that has a head

  • ffice in what city?

The Oberoi Group is a hotel company with its head office in Delhi.

Doc 2

Future Fibre Technologies a fiber technologies company …

Doc 3

The Oberoi family is an Indian family that is …

Doc 1

slide-42
SLIDE 42

Sentence Factored Model

The Oberoi family … what city? The Oberoi Group … in Delhi. Future Fibre Technologies is a fibre… The Oberoi family … Answer predic5on: Delhi

BiDAF BiDAF BiDAF

slide-43
SLIDE 43

Sentence Factored Model

The Oberoi family … what city? The Oberoi Group … in Delhi. Future Fibre Technologies is a fibre… The Oberoi family … Answer predic5on: Delhi

  • Sofmax over all sentences is the only cross-sentence interac5on

BiDAF BiDAF BiDAF

slide-44
SLIDE 44

BiDAF++ QFE GRN DFGN Sentence Factored

F1 35 70

50.8 69.7 69.0 68.1 58.7

Results on HotpotQA

BiDAF++ QFE GRN DFGN Sentence Factored

slide-45
SLIDE 45

BiDAF++ QFE GRN DFGN Sentence Factored

F1 35 70

50.8 69.7 69.0 68.1 58.7

Results on HotpotQA

BiDAF++ QFE GRN DFGN Sentence Factored

slide-46
SLIDE 46

BiDAF++ QFE GRN DFGN Sentence Factored

F1 35 70

50.8 69.7 69.0 68.1 58.7

Results on HotpotQA

BiDAF++ QFE GRN DFGN Sentence Factored

slide-47
SLIDE 47

BiDAF++ QFE GRN DFGN Sentence Factored

F1 35 70

50.8 69.7 69.0 68.1 58.7

Results on HotpotQA

BiDAF++ QFE GRN DFGN Sentence Factored

slide-48
SLIDE 48

BiDAF++ QFE GRN DFGN Sentence Factored

F1 35 70

50.8 69.7 69.0 68.1 58.7

A simple single sentence reasoning model could solve more than half ques5ons on HotpotQA.

Results on HotpotQA

BiDAF++ QFE GRN DFGN Sentence Factored

slide-49
SLIDE 49

Sentence Factored Model

The Oberoi family … what city? The Oberoi Group … in Delhi. Future Fibre Technologies … The Oberoi family … Containing answer: 1

BiDAF BiDAF BiDAF

slide-50
SLIDE 50

Sentence Factored Model

The Oberoi family … what city? The Oberoi Group … in Delhi. Future Fibre Technologies … The Oberoi family … Containing answer: 1

  • Predict which sentence contains the answer

BiDAF BiDAF BiDAF

slide-51
SLIDE 51

Sentence Factored Model

The Oberoi family … what city? The Oberoi Group … in Delhi. Future Fibre Technologies … The Oberoi family … Containing answer: 1

  • Predict which sentence contains the answer

BiDAF BiDAF BiDAF

  • A simpler way to test how much mul5-hop reasoning is needed
slide-52
SLIDE 52

Accuracy 45 90

5.4 57.2 6.5 66.1 22.1 88

SQuAD HotpotQA

Answer Sentence Predic5on Results

WikiHop SQuAD HotpotQA WikiHop

Random Sentence-factored

slide-53
SLIDE 53

Accuracy 45 90

5.4 57.2 6.5 66.1 22.1 88

SQuAD HotpotQA

Answer Sentence Predic5on Results

WikiHop SQuAD HotpotQA WikiHop

Random Sentence-factored

  • Answers can be found for

many ques5ons without a mul5-hop process.

slide-54
SLIDE 54

Research Ques5ons

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

Government posi5on Chief of Protocol, actress, singer Span-based vs. Mul5ple-choice Oberoi Family Delhi

slide-55
SLIDE 55

Making HotpotQA Mul5ple-Choice

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head

  • ffice in what city?

The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2

Future Fibre Technologies a fiber technologies company with its US head office in Mountain View, Middle East head office in Dubai …

Doc 3

slide-56
SLIDE 56

Making HotpotQA Mul5ple-Choice

The Oberoi family is an Indian family that is famous for its involvement in hotels, namely through The Oberoi Group Ques%on: The Oberoi family is part of a hotel company that has a head

  • ffice in what city?

The Oberoi Group is a hotel company with its head office in Delhi.

Doc 1 Doc 2

Future Fibre Technologies a fiber technologies company with its US head office in Mountain View, Middle East head office in Dubai …

Doc 3

  • Randomly choose en55es in

the documents as candidates. The Oberoi family, The Oberoi Group, Delhi, Future Fibre Technologies, US, Mountain View, Middle East, Dubai

slide-57
SLIDE 57

Results of HotpotQA Mul5ple-Choice

40 60 80 10 20 30 40 50

NoContext BiDAF++

Number of candidates Accuracy

slide-58
SLIDE 58

Results of HotpotQA Mul5ple-Choice

40 60 80 10 20 30 40 50

NoContext BiDAF++

Number of candidates Accuracy

2.0

slide-59
SLIDE 59

Results of HotpotQA Mul5ple-Choice

40 60 80 10 20 30 40 50

NoContext BiDAF++

Number of candidates Accuracy

2.0 2.9

slide-60
SLIDE 60

Results of HotpotQA Mul5ple-Choice

40 60 80 10 20 30 40 50

NoContext BiDAF++

Number of candidates Accuracy

2.0 2.9 3.8

slide-61
SLIDE 61

Results of HotpotQA Mul5ple-Choice

40 60 80 10 20 30 40 50

NoContext BiDAF++

Number of candidates Accuracy

2.0 2.9 3.8 2.5

slide-62
SLIDE 62

Results of HotpotQA Mul5ple-Choice

40 60 80 10 20 30 40 50

NoContext BiDAF++

Number of candidates Accuracy

2.0 2.9 3.8 2.5 2.5

slide-63
SLIDE 63

Results of HotpotQA Mul5ple-Choice

40 60 80 10 20 30 40 50

NoContext BiDAF++

More candidates does make the predic5on harder, but BiDAF++ does not learn more than the NoContext baseline

Number of candidates Accuracy

2.0 2.9 3.8 2.5 2.5

slide-64
SLIDE 64

Conclusions

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

slide-65
SLIDE 65

Conclusions

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

  • Models can achieve high performance by “chea5ng” through

extra candidate informa5on.

slide-66
SLIDE 66

Conclusions

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

  • Models can achieve high performance by “chea5ng” through

extra candidate informa5on.

  • More than half of the ques5ons in WikiHop and HotpotQA can

be answered in a single hop.

slide-67
SLIDE 67

Conclusions

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

  • Models can achieve high performance by “chea5ng” through

extra candidate informa5on.

  • More than half of the ques5ons in WikiHop and HotpotQA can

be answered in a single hop.

  • Span-based datasets may be less hackable.
slide-68
SLIDE 68

Conclusions

Can a model iden5fy the answer with only a set of candidates? Can a model iden5fy where the answer is in a single hop? Does the type of supervision ma;er?

  • Models can achieve high performance by “chea5ng” through

extra candidate informa5on.

  • More than half of the ques5ons in WikiHop and HotpotQA can

be answered in a single hop.

  • Span-based datasets may be less hackable.

*No results for WikiHop-Masked or HotpotQA fullwiki sepngs

slide-69
SLIDE 69

Takeaways

When construc5ng mul5-hop based dataset:

slide-70
SLIDE 70

Takeaways

  • If using mul5ple-choice supervision, check the no-context baseline to see

whether there are strong correla5ons between ques5on and candidates. When construc5ng mul5-hop based dataset:

slide-71
SLIDE 71

Takeaways

  • If using mul5ple-choice supervision, check the no-context baseline to see

whether there are strong correla5ons between ques5on and candidates.

  • Do an adversarial test using a sentence-factored model.

When construc5ng mul5-hop based dataset:

slide-72
SLIDE 72

Takeaways

  • If using mul5ple-choice supervision, check the no-context baseline to see

whether there are strong correla5ons between ques5on and candidates.

  • Do an adversarial test using a sentence-factored model.

When construc5ng mul5-hop based dataset:

Thank you! Q & A