Yaphets24 8977be1df3 [Bugfix]Fix deepseek 3.2 C8 precision by rotary tensor (#7537) ...

### What this PR does / why we need it?
During the attention quantization process of DeepSeek V3.2, it is
necessary to retrieve the Hadamard matrix from the weights to facilitate
the computation.

### Does this PR introduce _any_ user-facing change?
No. But there will be two new tensor in quant weight.

### How was this patch tested?

- vLLM version: v0.18.0
- vLLM main:
8b6325758c

---------

Signed-off-by: mayumeng <m30059191@china.huawei.com>
Co-authored-by: mayumeng <m30059191@china.huawei.com>

2026-03-25 09:18:00 +08:00

..

methods

[Bugfix]Fix deepseek 3.2 C8 precision by rotary tensor (#7537)

2026-03-25 09:18:00 +08:00

__init__.py

[refactor] replace scattered business kwargs with typed request objects and explicit stage boundaries (#7024)

2026-03-20 23:23:57 +08:00

compressed_tensors_config.py

[CI] Add pre-commit check for patch logger (#7446)

2026-03-19 16:53:20 +08:00

method_adapters.py

[refactor] replace scattered business kwargs with typed request objects and explicit stage boundaries (#7024)

2026-03-20 23:23:57 +08:00

modelslim_config.py

[Bugfix]Fix deepseek 3.2 C8 precision by rotary tensor (#7537)

2026-03-25 09:18:00 +08:00

quant_parser.py

[misc] move mxfp_compat into device to decouple from quantization init chain (#6918)

2026-03-02 18:17:01 +08:00

quant_type.py

[refactor] replace scattered business kwargs with typed request objects and explicit stage boundaries (#7024)

2026-03-20 23:23:57 +08:00

utils.py

[CI] Add pre-commit check for patch logger (#7446)

2026-03-19 16:53:20 +08:00