modelscope / modelscope/ms-swift
为什么微调qwen3.5-9B模型后FP8量化vllm部署，和不量化的模型的输出差距非常大，感觉训练的内容遗忘很多
Open
Nobody has claimed this yet.
question
- Dominant language
- Python
- Stars
- 15.7k
- Forks
- 1.7k
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 136
Description
Checklist / 检查清单
- I have searched existing issues, and this is a new question or discussion topic. / 我已经搜索过现有的 issues，确认这是一个新的问题与讨论。
Question Description / 问题描述
微调的内容遗忘很多，之前qwen3-8B的时候fp8量化都比较正常，至少训练的内容例如一些输出json格式能力还是基本ok的，量化qwen3.5-9B模型后根据不按照格式输出
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The report names no repository files, tests, or entry points. Start by collecting the fine-tuning configuration, FP8 quantization command, vLLM deployment command, and comparable outputs for the unquantized model; done means determining whether the difference is expected, reproducible, or requires a documented fix.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100