REVIEW 10 cited by
Is Flash Attention Stable?
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
Is Flash Attention Stable?
read the original abstract
Training large-scale machine learning models poses distinct system challenges, given both the size and complexity of today's workloads. Recently, many organizations training state-of-the-art Generative AI models have reported cases of instability during training, often taking the form of loss spikes. Numeric deviation has emerged as a potential cause of this training instability, although quantifying this is especially challenging given the costly nature of training runs. In this work, we develop a principled approach to understanding the effects of numeric deviation, and construct proxies to put observations into context when downstream effects are difficult to quantify. As a case study, we apply this framework to analyze the widely-adopted Flash Attention optimization. We find that Flash Attention sees roughly an order of magnitude more numeric deviation as compared to Baseline Attention at BF16 when measured during an isolated forward pass. We then use a data-driven analysis based on the Wasserstein Distance to provide upper bounds on how this numeric deviation impacts model weights during training, finding that the numerical deviation present in Flash Attention is 2-5 times less significant than low-precision training.
Forward citations
Cited by 10 Pith papers
-
Fingerprinting Inference Systems of Large Language Models
Inference system components of LLMs can be fingerprinted from observable prompt-response behavior due to characteristic numerical deviations.
-
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.
-
FlashAttention-3: Fast and Accurate Attention with Asynchrony and Low-precision
FlashAttention-3 achieves 1.5-2x speedup on H100 GPUs for attention, reaching 740 TFLOPs/s (75% utilization) in FP16 and near 1.2 PFLOPs/s in FP8 while cutting numerical error by 2.6x versus baseline FP8 attention.
-
Reference Traces for Auditing Invisible Weight Updates and Guiding Exact-Budget Protection
Weight updates that fall below half a ULP freeze coordinates deterministically, and freeze time is predictable a priori from a high-precision trajectory and mantissa length alone.
-
FastTPS: An Optimized Method for LLM Token Phase for AI accelerators
FastTPS accelerates LLM token-phase inference via reloading-free static KV-cache management, tiled fused RoPE attention, and interlaced-weight MLP fusion, yielding up to 6× speedup at 93% bandwidth on AMD NPUs.
-
An Efficient vLLM-Based Inference Pipeline for Unified Audio Understanding and Generation
Extends vLLM with delay-pattern de-interleaving, multi-stream sampling, and co-scheduled CFG to achieve 80% of non-CFG throughput for unified audio tasks while open-sourcing the pipeline.
-
Gated KalmaNet: A Fading Memory Layer Through Test-Time Ridge Regression
Gated KalmaNet uses exact Kalman gain computation with adaptive gating and Chebyshev iteration to improve SSM performance on long-context tasks over prior approximations like DeltaNet.
-
Reference Traces for Auditing Invisible Weight Updates and Guiding Exact-Budget Protection
A matched high-precision reference trace forecasts invisible weight-update events in controlled low-precision runs, and a pre-frozen 10% fp32-master mask recovers ~61% of the RNE-vs-fp32 loss gap on three unseen targets.
-
PRISM: Probabilistic Runtime Insights and Scalable Performance Modeling for Large-Scale Distributed Training
PRISM introduces a probabilistic performance modeling framework that quantifies guarantees on training time for large-scale distributed systems under runtime variability.
-
A Survey on Efficient Inference for Large Language Models
The paper surveys techniques to speed up and reduce the resource needs of LLM inference, organized by data-level, model-level, and system-level changes, with comparative experiments on representative methods.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.