Low-precision Flash Attention fails due to similar low-rank attention representations combined with biased rounding errors that accumulate and corrupt weight updates; a minimal fix to reduce rounding bias stabilizes training.
Low-precision training of large language models: Methods, challenges, and opportunities.arXiv preprint arXiv:2505.01043
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
StoSignSGD resolves SignSGD divergence on non-smooth objectives via structural stochasticity, matching optimal convex rates and improving non-convex bounds while delivering 1.44-2.14x speedups in FP8 LLM pretraining.
Proposes High-Precision Scoring (HPS) and Tie-aware Retrieval Metrics (TRM) to reduce tie-induced instability in low-precision retrieval evaluation.
citing papers explorer
-
Why Low-Precision Transformer Training Fails: An Analysis on Flash Attention
Low-precision Flash Attention fails due to similar low-rank attention representations combined with biased rounding errors that accumulate and corrupt weight updates; a minimal fix to reduce rounding bias stabilizes training.
-
StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models
StoSignSGD resolves SignSGD divergence on non-smooth objectives via structural stochasticity, matching optimal convex rates and improving non-convex bounds while delivering 1.44-2.14x speedups in FP8 LLM pretraining.
-
Reliable Evaluation Protocol for Low-Precision Retrieval
Proposes High-Precision Scoring (HPS) and Tie-aware Retrieval Metrics (TRM) to reduce tie-induced instability in low-precision retrieval evaluation.
- Decomposing MXFP4 quantization error for LLM reinforcement learning: reducible bias, recoverable deadzone, and an irreducible floor