finding
active
finding:for-six-of-eight-tasks-all-but-aqua-and-logiqa-the-13b-model-gives-the-most-faithful-reasoning-better-than-smaller-modelsFor six of eight tasks (all but AQuA and LogiQA) the 13B model gives the most faithful reasoning, better than smaller models
Optimum at mid-size.
Source paper
extracted_from(2023) · Tamera Lanham · Anna Chen · Ansh Radhakrishnan · Benoit Steiner +26
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (2)
claim
- Interpretation of §3.
- Rejecting alternative explanation.
Questions (1)
question
- Motivates §3.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- AQuA and LogiQA show the most faithful reasoning on both early answering and adding mistakes by some marginfinding0.826Top two tasks.
- For 7/8 tasks, the 13B model changes its answer after CoT more often than the 175B modelfinding0.811Inverse scaling.
- Explanation supported by §3.
- Appendix C.
- Appendix C.
- Larger models linearly represent more general concepts including truth
- Practical implication.
- Inverse scaling interpretation.