Update vllm version to support llama3.1 (#705)
This commit is contained in:
@@ -73,6 +73,8 @@ def get_context_length(config):
|
||||
rope_scaling = getattr(config, "rope_scaling", None)
|
||||
if rope_scaling:
|
||||
rope_scaling_factor = config.rope_scaling["factor"]
|
||||
if config.rope_scaling["rope_type"] == "llama3":
|
||||
rope_scaling_factor = 1
|
||||
else:
|
||||
rope_scaling_factor = 1
|
||||
|
||||
|
||||
Reference in New Issue
Block a user