Use hf_pretrained_model_name_or_path for tokenizer

abetlen · abetlen · commit a79743b15764 · 2024-01-31T16:08:47.000-05:00
diff --git a/llama_cpp/server/model.py b/llama_cpp/server/model.py
@@ -92,7 +92,7 @@ def load_llama_from_model_settings(settings: ModelSettings) -> llama_cpp.Llama:
                     json.load(open(settings.hf_tokenizer_config_path))
                 )
             )
-        
+
         tokenizer: Optional[llama_cpp.BaseLlamaTokenizer] = None
         if settings.hf_pretrained_model_name_or_path is not None:
             tokenizer = llama_cpp.LlamaHFTokenizer.from_pretrained(settings.hf_pretrained_model_name_or_path)
@@ -130,8 +130,6 @@ def load_llama_from_model_settings(settings: ModelSettings) -> llama_cpp.Llama:
             use_mmap=settings.use_mmap,
             use_mlock=settings.use_mlock,
             kv_overrides=kv_overrides,
-            # Tokenizer Params (optionally for Functionary function calling)
-            hf_tokenizer_path=settings.hf_tokenizer_path,
             # Context Params
             seed=settings.seed,
             n_ctx=settings.n_ctx,
diff --git a/llama_cpp/server/settings.py b/llama_cpp/server/settings.py
@@ -56,11 +56,6 @@ class ModelSettings(BaseSettings):
         default=None,
         description="List of model kv overrides in the format key=type:value where type is one of (bool, int, float). Valid true values are (true, TRUE, 1), otherwise false.",
     )
-    # Tokenizer Params
-    hf_tokenizer_path: Optional[str] = Field(
-        default=None,
-        description="Override llama.cpp tokenizer with HF AutoTokenizer from this path if provided.",
-    )
     # Context Params
     seed: int = Field(
         default=llama_cpp.LLAMA_DEFAULT_SEED, description="Random seed. -1 for random."
