[Fix] Enable mm_processor_cache with vision LoRA (#31927)
Signed-off-by: prashanth058 <prashanth.dannamaneni@uipath.com>
This commit is contained in:
@@ -1649,19 +1649,6 @@ class EngineArgs:
|
||||
else None
|
||||
)
|
||||
|
||||
if (
|
||||
lora_config is not None
|
||||
and lora_config.enable_tower_connector_lora
|
||||
and self.mm_processor_cache_gb != 0
|
||||
):
|
||||
raise ValueError(
|
||||
"Currently, enable_tower_connector_lora is "
|
||||
"incompatible with the multi-modal processor cache. "
|
||||
"When enable_tower_connector_lora is set, "
|
||||
"mm_processor_cache_gb must be 0, got %s",
|
||||
self.mm_processor_cache_gb,
|
||||
)
|
||||
|
||||
if (
|
||||
lora_config is not None
|
||||
and speculative_config is not None
|
||||
|
||||
Reference in New Issue
Block a user