REVIEW 4 cited by
Measuring Causal Effects of Data Statistics on Language Model's `Factual' Predictions
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Large amounts of training data are one of the major reasons for the high performance of state-of-the-art NLP models. But what exactly in the training data causes a model to make a certain prediction? We seek to answer this question by providing a language for describing how training data influences predictions, through a causal framework. Importantly, our framework bypasses the need to retrain expensive models and allows us to estimate causal effects based on observational data alone. Addressing the problem of extracting factual knowledge from pretrained language models (PLMs), we focus on simple data statistics such as co-occurrence counts and show that these statistics do influence the predictions of PLMs, suggesting that such models rely on shallow heuristics. Our causal framework and our results demonstrate the importance of studying datasets and the benefits of causality for understanding NLP models.
Forward citations
Cited by 4 Pith papers
-
When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models
LLM first-answer accuracy on procedural arithmetic drops from 61% on 5-step tasks to 20% on 95-step tasks, with frequent failures including skipped steps, premature answers, and hallucinated operations.
-
Tracing LLM Behavior to the Training Data with Empirical Next-Token Distributions
A trained transformer's next-token distribution often matches the empirical next-token distribution of its pretraining corpus, with agreement improving as models grow, while a persistent tail of mismatches remains.
-
The Common Pile v0.1: An 8TB Dataset of Public Domain and Openly Licensed Text
A new 8TB openly-licensed text corpus trains 7B LLMs that are competitive with Llama 1/2, showing that performant models need not depend on unlicensed web data.
-
Protoknowledge Shapes Behaviour of LLMs in Downstream Tasks: Memorization and Generalization with Knowledge Graphs
A new 'protoknowledge' analysis links lexical, hierarchical, and topological knowledge graph recall in LLMs to Text-to-SPARQL performance, offering a black-box route to detect semantic contamination.
Discussion (0). Continue with ORCID to comment.