[bugfix] fix deepseek v4 fp8

[Jintao-Huang]

No description provided.

[gemini-code-assist]

Code Review

This pull request modifies deepseek_v4.py by directly accessing config.fp8_param instead of using getattr and by casting concatenated weights to torch.float8_e4m3fn when setting wo_a.weight. The reviewer raised two important concerns: first, removing getattr when checking fp8_param can lead to an AttributeError if standard configurations are passed; second, unconditionally calling .view(torch.float8_e4m3fn) on the concatenated weights can cause failures or data corruption if the parameters are not FP8, suggesting a guard check on the tensor's data type.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.