Add DeepSeekV4 fused MoE Triton autotune support (#25569)

Co-authored-by: undefined <zhouchen.arrebol@jd.com>
Co-authored-by: xq25478 <xq25478@qq.com>
Co-authored-by: xieminghe.simon <xieminghe.simon@jd.com>
This commit is contained in:
xieminghe1
2026-05-18 21:35:32 +08:00
committed by GitHub
co-authored by undefined xq25478 xieminghe.simon
parent f04c522534
commit ba2ffcf156
2 changed files with 6 additions and 0 deletions
@@ -85,6 +85,7 @@ def get_model_config(
"DeepseekV2ForCausalLM",
"DeepseekV3ForCausalLM",
"DeepseekV32ForCausalLM",
"DeepseekV4ForCausalLM",
"Glm4MoeForCausalLM",
"GlmMoeDsaForCausalLM",
"MistralLarge3ForCausalLM",