[model] Support fp8 with fuse model

[mgilmore-relace]

What does this PR do?

Add concise overview of what this PR aims to achieve or accomplish. Reference related GitHub issues and PRs that help with the review.

Checklist Before Starting

• Search for similar PRs. Paste at least one query link here: ...
• Format the PR title as [{modules}] {type}: {description} (This will be checked by the CI)
  • {modules} include fsdp, megatron, sglang, vllm, rollout, trainer, ci, training_utils, recipe, hardware, deployment, ray, worker, single_controller, misc, perf, model, algo, env, tool, ckpt, doc, data, cfg, reward
  • If this PR involves multiple modules, separate them with , like [megatron, fsdp, doc]
  • {type} is in feat, fix, refactor, chore, test
  • If this PR breaks any API (CLI arguments, config, function signature, etc.), add [BREAKING] to the beginning of the title.
  • Example: [BREAKING][fsdp, megatron] feat: dynamic batching

Test

For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc.

API and Usage Example

Demonstrate how the API changes if any, and provide usage example(s) if possible.

# Add code snippet or script demonstrating how to use this

Design & Code Changes

Demonstrate the high-level design if this PR is complex, and list the specific changes.

Checklist Before Submitting

Important

Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review.

• Read the Contribute Guide.
• Apply pre-commit checks: pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always
• Add / Update the documentation.
• Add unit or end-to-end test(s) to the CI workflow to cover all the code. If not feasible, explain why: ...
• Once your PR is ready for CI, send a message in the ci-request channel in the verl Slack workspace. (If not accessible, please try the Feishu group (飞书群).)

[gemini-code-assist]

Code Review

This pull request introduces support for FP8 quantization in the fused model forward pass. The changes involve detecting if FP8 is enabled via the model configuration and then passing a flag to preprocess_packed_seqs to trigger special padding logic required for FP8 performance and correctness. My review identified a potential issue where the logic to detect if FP8 padding should be used might not work correctly with the documented configuration, which could prevent the feature from functioning as intended. I've provided a suggestion to make this check more robust.

[gemini-code-assist]

The condition to enable FP8 padding appears to be inconsistent with the project's documentation. According to docs/advance/fp8.md, FP8 quantization is enabled by setting quantization: "fp8". However, the current code checks if fp8 in ["e4m3", "hybrid"]. If the configuration value for fp8 is the string "fp8", use_fp8_padding will be False, and the necessary padding for FP8-optimized kernels will not be applied. This could lead to incorrect behavior or performance issues.

To align with the documentation and make the feature work as expected, the condition should be updated to correctly handle the "fp8" configuration value.

Suggested change

	use_fp8_padding = fp8 in ["e4m3", "hybrid"]
	use_fp8_padding = bool(fp8)