Clean up wrapper in flashinfer backend (#2638)

2024-12-29 00:45:57 -08:00
parent fd34f2da35
commit 3815b23ccb
12 changed files with 197 additions and 94 deletions
--- a/python/sglang/srt/configs/model_config.py
+++ b/python/sglang/srt/configs/model_config.py
@@ -131,10 +131,8 @@ class ModelConfig:
        # Veirfy quantization
        self._verify_quantization()

-        # Text attrs
+        # Cache attributes
        self.hf_eos_token_id = self.get_hf_eos_token_id()
-
-        # Multimodel attrs
        self.image_token_id = getattr(self.hf_config, "image_token_id", None)

    # adapted from https://github.com/vllm-project/vllm/blob/main/vllm/config.py#L289