Model: Adjust draft_gpu_split and add to config

The previous code overrode the existing gpu split and device idx
values. This now sets an independent draft_gpu_split value and
adjusts the gpu_devices check only if the draft_gpu_split array
is larger than the gpu_split array.

Draft gpu split is not Tensor Parallel, and defaults to gpu_split_auto
if a split is not provided.

Signed-off-by: kingbri <8082010+kingbri1@users.noreply.github.com>

common/config_models.py

@@ -351,6 +351,13 @@ class DraftModelConfig(BaseConfigModel):
             f"Possible values: {str(CACHE_SIZES)[15:-1]}."
         ),
     )
+    draft_gpu_split: List[float] = Field(
+        default_factory=list,
+        description=(
+            "An integer array of GBs of VRAM to split between GPUs (default: []).\n"
+            "If this isn't filled in, the draft model is autosplit."
+        ),
+    )
 
 
 class LoraInstanceModel(BaseConfigModel):