vllm-project/vllm

[Transformers v5] SarvamMLAForCausalLM

Open

#38,734 创建于 2026年4月1日

在 GitHub 查看
 (2 评论) (0 反应) (1 负责人)Python (80,034 star) (16,816 fork)batch import
good first issuehelp wanted

描述

This is a sub-issue forming part of the work in https://github.com/vllm-project/vllm/issues/38379, please read the description of this issue before beginning to work on this one.

Which test is failing?

As of https://github.com/huggingface/transformers/pull/41250 the ignore_keys argument to validate_rope was removed in favour of ClassVars (ignore_keys_at_rope_validation) attached to the config classes themselves.

$ pytest tests/models/test_initialization.py::test_can_initialize_large_subset[SarvamMLAForCausalLM]
...
TypeError: RotaryEmbeddingConfigMixin.validate_rope() got an unexpected keyword argument 'ignore_keys'

How to configure my environment?

It's very important that you install both vLLM and Transformers from source so that your test results reflect the current state of both libraries.

# Or your fork
git clone https://github.com/huggingface/transformers.git
git clone https://github.com/vllm-project/vllm.git

cd vllm
VLLM_USE_PRECOMPILED=1 uv pip install -e .
uv pip install -e ../transformers

贡献者指南

技术栈
python
领域
backendmachine learningtesting
议题类型
bug
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
3
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
1-3 hours
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
active
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
Basic knowledge of PythonFamiliarity with vLLM and TransformersUnderstanding of validate rope method
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
65
研究方向
First, read the linked issue #38379 for context. Then examine the PR #41250 that removed the `ignore keys` argument from `validate rope`. Update the test `test can initialize large subset` for SarvamMLAForCausalLM to work with the new signature. Ensure that the `ignore keys at rope validation` ClassVar is properly set in the config class for SarvamMLA. Run the test using the instructions provided.