MétaCan
Menu
Back to cohort
Record W4287122359 · doi:10.48550/arxiv.2106.05346

End-to-End Training of Multi-Document Reader and Retriever for\n Open-Domain Question Answering

2021· preprint· W4287122359 on OpenAlex

Why this work is in the frame

A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.

aboutThe title or abstract carries a Canadian signal from the geographic lexicon.
no affNo Canadian affiliation: this work is invisible to an affiliation-only frame.
No Canadian affiliation. An affiliation-only frame, the usual design, would never have seen this work. It is one of the works that make the case for inverting the frame.

Bibliographic record

VenuearXiv (Cornell University) · 2021
Typepreprint
Language
FieldComputer Science
TopicTopic Modeling
Canadian institutionsnot available
Fundersnot available
KeywordsComputer scienceQuestion answeringBenchmark (surveying)Information retrievalDomain (mathematical analysis)Set (abstract data type)Training setOpen domainArtificial intelligenceLabrador RetrieverRelevance (law)Machine learningMathematics

Abstract

fetched live from OpenAlex

We present an end-to-end differentiable training method for\nretrieval-augmented open-domain question answering systems that combine\ninformation from multiple retrieved documents when generating answers. We model\nretrieval decisions as latent variables over sets of relevant documents. Since\nmarginalizing over sets of retrieved documents is computationally hard, we\napproximate this using an expectation-maximization algorithm. We iteratively\nestimate the value of our latent variable (the set of relevant documents for a\ngiven question) and then use this estimate to update the retriever and reader\nparameters. We hypothesize that such end-to-end training allows training\nsignals to flow to the reader and then to the retriever better than staged-wise\ntraining. This results in a retriever that is able to select more relevant\ndocuments for a question and a reader that is trained on more accurate\ndocuments to generate an answer. Experiments on three benchmark datasets\ndemonstrate that our proposed method outperforms all existing approaches of\ncomparable size by 2-3% absolute exact match points, achieving new\nstate-of-the-art results. Our results also demonstrate the feasibility of\nlearning to retrieve to improve answer generation without explicit supervision\nof retrieval decisions.\n

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

Full frame distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.002
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMeta-epidemiology (narrow)
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: Simulation or modeling
GenreCandidate signal: Empirical · Consensus signal: none
Teacher disagreement score0.475
Threshold uncertainty score0.999

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0020.000
Meta-epidemiology (narrow)0.0010.001
Meta-epidemiology (broad)0.0010.000
Bibliometrics0.0000.001
Science and technology studies0.0000.000
Scholarly communication0.0010.001
Open science0.0020.006
Research integrity0.0000.001
Insufficient payload (model declined to judge)0.0000.000

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.148
GPT teacher head0.244
Teacher spread0.096 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it