Research map / Large language models / Language, safety and interpretability
Sequences of Logits Reveal the Low Rank Structure of Language Models
Noah Golowich, Allen Liu, Abhishek Shetty
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| train the model (vs. untrained step-0 checkpoint) (ablation) | OLMo-1b · wiki logit matrix | singular value power-law exponent: 0.374 -> 0.561 | Improved |
| use extended logit matrix (multi-token futures) instead of single-token logit matrix for generation (ablation) | LINGEN · OLMo-1b generation | KL divergence to true model | Improved |
| generate using linear combination of model outputs on in-distribution histories | LINGEN · OLMo-1b, wiki | KL divergence per token | Improved |
| generate using nonsense histories and futures | LINGEN · OLMo-1b, out-of-distribution | KL divergence per token | Mixed |
| compare against earlier training checkpoint baseline | LINGEN · OLMo-1b | KL divergence | Improved |
| increase k (columns) from 50 to 200 or choose random tokens (ablation) | logit matrix · OLMo-1b | power-law exponent: 0.561 -> 0.543 / 0.527 | No clear effect |
| scale up size of histories and futures (ablation) | logit matrix · OLMo-1b | power-law exponent alpha | No clear effect |
| project futures to random-token (nonsense) futures | column space of low-rank approximation · OLMo-1b | principal angle cosines | No clear effect |
Closest papers by meaning
- Weak-to-Strong Search: Align Large Language Models via Searching over Small Language Models (NeurIPS 2024 · similarity 0.87)
- On the Worst Prompt Performance of Large Language Models (NeurIPS 2024 · similarity 0.87)
- A Probabilistic Perspective on Unlearning and Alignment for Large Language Models (ICLR 2025 · similarity 0.87)
- The Truth is in There: Improving Reasoning in Language Models with Layer-Selective Rank Reduction (ICLR 2024 · similarity 0.86)
- Order-Independence Without Fine Tuning (NeurIPS 2024 · similarity 0.86)
- Lines of Thought in Large Language Models (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
