Float8 GRPO, RL #3640
Float8 GRPO, RL #3640
Conversation
* Enable FP8 + RL training for bf16 models **Summary:** Enable FP8 + RL training using TorchAO for 1.33x faster training and 42% less model memory usage: - We quantize the frozen LoRA weights into fp8 and keep the LoRA adapters in bf16 - We leverage TorchAO's `Float8Tensor`, which calls into fbgemm's fp8 x fp8 rowwise matmul kernel - For now, we need to do an offline quantization first, because vllm doesn't support on-the-fly quantization for torchao yet (this is in progress: vllm-project/vllm#26327) **Example usage:** ``` model, tokenizer = FastLanguageModel.from_pretrained( model_name = "unsloth/Qwen3-8B-Base", max_seq_length = 2048, load_in_4bit = False, fast_inference = True, max_lora_rank = 32, load_in_fp8 = True, # set this to True ) \# the rest is the same as before model = FastLanguageModel.get_peft_model(...) ``` **Initial results:** ``` \# fp8 {'train_runtime': 1725.4337, 'train_samples_per_second': 0.232, 'train_steps_per_second': 0.058, 'train_loss': 0.00015715716748673002, 'epoch': 0.01} \# bf16 {'train_runtime': 2297.8145, 'train_samples_per_second': 0.174, 'train_steps_per_second': 0.044, 'train_loss': 0.00016081033063528594, 'epoch': 0.01} ``` <img width="1199" height="448" alt="Screenshot 2025-11-11 at 4 10 50 PM" src="https://github.com/user-attachments/assets/b6304afd-89e9-42b1-8064-775807e17b23" /> Test script: https://gist.github.com/andrewor14/5b85119fae46845d07b608d420907423 **Requires:** - pytorch/ao#3158 (torchao nightly or 0.15.0+) - unslothai/unsloth-zoo#351 * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update utils.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * _get_inference_mode_context_manager * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update utils.py * Update utils.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: Daniel Han <danielhanchen@gmail.com>
* make loading gpt-oss-BF16 faster. Linked to unsloth-zoo PR #314 * fix model loading and clean merged model directory * revert default quant * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * revert mapper.py --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
* Add 128x128 PerBlock FP8 + RL **Summary:** Following #3440, this PR extends torchao FP8 + RL support to also handle 128x128 PerBlock granularity (in addition to PerRow). **Example usage:** ``` model, tokenizer = FastLanguageModel.from_pretrained( model_name = "unsloth/Qwen3-8B-Base", max_seq_length = 2048, load_in_4bit = False, fast_inference = True, max_lora_rank = 32, load_in_fp8 = "block", # or "row" or True ) ``` **Initial results:** TBD **Note:** - Requires pytorch/ao#3370 * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
for more information, see https://pre-commit.ci
Summary of ChangesHello @danielhanchen, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly expands the project's capabilities by introducing robust support for Float8 (FP8) quantization, which can lead to substantial memory savings and performance gains, especially on newer GPUs. It also updates the project's compatibility with the latest PyTorch and CUDA versions, refines the model loading process to seamlessly integrate FP8, and includes several key stability and integration improvements across various modules, including RL training and dependency management. Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
![gemini-code-assist[bot]](https://cdn.statically.io/img/avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
The pull request introduces comprehensive support for FP8 quantization using torchao and fbgemm_gpu, enhancing performance and memory efficiency. Key changes include dynamic compatibility checks for FBGEMM, a robust workaround for a PyTorch inference mode bug with torchao quantized models, and updated dependency management. The changes are well-integrated across various modules, from installation scripts and pyproject.toml to core kernel functions and model loading utilities. However, there is some code duplication in the model loading logic that could be refactored for better maintainability. Additionally, some error messages and TODOs indicate areas for further investigation and refinement.
No description provided.