Tree: Remove fasttensors

Now a noop in upstream.

Signed-off-by: kingbri <[email protected]>

backends/exllamav2/model.py

@@ -251,9 +251,6 @@ async def create(cls, model_directory: pathlib.Path, quiet=False, **kwargs):
         else:
             self.config.scale_alpha_value = rope_alpha
 
-        # Enable fasttensors loading if present
-        self.config.fasttensors = unwrap(kwargs.get("fasttensors"), False)
-
         # Set max batch size to the config override
         self.max_batch_size = unwrap(kwargs.get("max_batch_size"))
 

common/config_models.py

@@ -290,13 +290,6 @@ class ModelConfig(BaseConfigModel):
         ),
         ge=1,
     )
-    fasttensors: Optional[bool] = Field(
-        False,
-        description=(
-            "Enables fasttensors to possibly increase model loading speeds "
-            "(default: False)."
-        ),
-    )
 
     _metadata: Metadata = PrivateAttr(Metadata())
     model_config = ConfigDict(protected_namespaces=())

config_sample.yml

@@ -135,9 +135,6 @@ model:
   # WARNING: Don't set this unless you know what you're doing!
   num_experts_per_token:
 
-  # Enables fasttensors to possibly increase model loading speeds (default: False).
-  fasttensors: false
-
 # Options for draft models (speculative decoding)
 # This will use more VRAM!
 draft_model:

endpoints/core/types/model.py

@@ -106,7 +106,6 @@ class ModelLoadRequest(BaseModel):
     chunk_size: Optional[int] = None
     prompt_template: Optional[str] = None
     num_experts_per_token: Optional[int] = None
-    fasttensors: Optional[bool] = None
 
     # Non-config arguments
     draft: Optional[DraftModelLoadRequest] = None