Pith. sign in

REVIEW 14 cited by

FP8-LM: Training FP8 Large Language Models

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

arxiv 2310.18313 v2 pith:55FVKBP5 submitted 2023-10-27 cs.LG cs.CL

classification cs.LGcs.CL
keywords trainingframeworkmixed-precisionlargellmsmodelsdatadistributed
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

In this paper, we explore FP8 low-bit data formats for efficient training of large language models (LLMs). Our key insight is that most variables, such as gradients and optimizer states, in LLM training can employ low-precision data formats without compromising model accuracy and requiring no changes to hyper-parameters. Specifically, we propose a new FP8 automatic mixed-precision framework for training LLMs. This framework offers three levels of FP8 utilization to streamline mixed-precision and distributed parallel training for LLMs. It gradually incorporates 8-bit gradients, optimizer states, and distributed learning in an incremental manner. Experiment results show that, during the training of GPT-175B model on H100 GPU platform, our FP8 mixed-precision training framework not only achieved a remarkable 39% reduction in real memory usage but also ran 75% faster than the widely adopted BF16 framework (i.e., Megatron-LM), surpassing the speed of Nvidia Transformer Engine by 37%. This largely reduces the training costs for large foundation models. Furthermore, our FP8 mixed-precision training methodology is generic. It can be seamlessly applied to other tasks such as LLM instruction tuning and reinforcement learning with human feedback, offering savings in fine-tuning expenses. Our FP8 low-precision training framework is open-sourced at {https://github.com/Azure/MS-AMP}{aka.ms/MS.AMP}.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 14 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. FP4 All the Way: Fully Quantized Training of LLMs

    cs.LG 2025-05 conditional novelty 7.0 of 10

    A 7-billion parameter LLM trained almost entirely in 4-bit float precision matches a standard 16-bit baseline after a brief quantization-aware fine-tuning phase.

  2. GIFT: Geometry-Informed Low-precision Gradient Communication for LLM Pretraining

    cs.DC 2026-07 conditional novelty 6.0 of 10

    Transforming gradients into K-FAC-based coordinates before FP8 quantization reduces communication error and improves downstream task preservation over Euclidean FP8, with a 7.6% end-to-end speedup on 64 GH200 GPUs.

  3. DynamiQ: Accelerating Gradient Synchronization using Compressed Multi-hop All-reduce

    cs.LG 2026-02 conditional novelty 6.0 of 10

    DynamiQ compresses gradients for multi-hop all-reduce with per-block bit-width allocation, block reordering, and fused recompression, matching BF16 accuracy at ~5 bits per coordinate and improving time-to-accuracy by ...

  4. Compute Requirements for Algorithmic Innovation in Frontier AI Models

    cs.LG 2025-07 conditional novelty 6.0 of 10

    Estimated development compute for 36 LLM pretraining innovations shows half would remain possible under GPT-2-level or 8-H100 compute caps.

  5. Log-Normal Multiplicative Dynamics for Stable Low-Precision Training of Large Networks

    cs.LG 2025-06 conditional novelty 6.0 of 10

    LMD, a log-normal multiplicative-weight optimizer, trains ViT and GPT-2 from scratch and keeps accuracy under MXFP6 forward-pass quantization.

  6. Thunder-LLM: Efficiently Adapting LLMs to Korean with Minimal Resources

    cs.CL 2025-06 conditional novelty 6.0 of 10

    A cost-effective recipe consisting of tokenizer extension, continual pretraining, FP8 training, and SFT/DPO post-training yields Korean-English bilingual 8B models with top Korean benchmark scores.

  7. Scaling Law for Quantization-Aware Training

    cs.LG 2025-05 conditional novelty 6.0 of 10

    A unified QAT scaling law predicts 4-bit quantization error from model size, training tokens, and group size, showing activation outliers in the FC2 layer are the main W4A4 bottleneck.

  8. Gaussian Weight Sampling for Scalable, Efficient and Stable Pseudo-Quantization Training

    cs.LG 2025-05 conditional novelty 6.0 of 10

    Using a rounded Gaussian noise distribution for pseudo-quantization makes low-precision FP weight training stable and cheap, matching or approaching BF16 baseline loss in LLM pretraining.

  9. QuantSpec: Self-Speculative Decoding with Hierarchical Quantized KV Cache

    cs.LG 2025-02 conditional novelty 6.0 of 10

    QuantSpec accelerates long-context LLM decoding by giving the self-draft model a 4-bit hierarchical KV cache shared with an INT8-quality target cache, achieving up to ~2.5x speedup and mostly >90% acceptance.

  10. A Comprehensive FP8 Training Recipe for Reasoning-Enhanced Language Models

    cs.CL 2025-09 conditional novelty 4.0 of 10

    A hybrid-granularity FP8 training recipe achieves near-lossless accuracy versus BF16 on reasoning benchmarks with up to 22% faster training and 14% lower memory.

  11. Foundation Models for Clean Energy Forecasting: A Comprehensive Review

    eess.SY 2025-07 conditional novelty 3.0 of 10

    A survey of foundation model methods, data, and open problems for renewable energy forecasting, built from roughly 218 cited works.

  12. Low-Precision Training of Large Language Models: Methods, Challenges, and Opportunities

    cs.LG 2025-05 conditional novelty 2.0 of 10

    A review that organizes low-precision LLM training research into fixed-point, floating-point, and custom numeric format categories and discusses open challenges.

  13. DeepSeek: Paradigm Shifts and Technical Evolution in Large AI Models

    cs.AI 2025-07 unverdicted

    A survey of DeepSeek's V3 and R1 models covering MLA, MoE, MTP, GRPO, and training engineering, with no new experimental results.

  14. Resource-Efficient Language Models: Quantization for Fast and Accessible Inference

    cs.AI 2025-05 unverdicted

    A survey of post-training quantization techniques for large language models, covering schemes, granularities, and popular methods, with no new experimental results.

Pith tools