Check eagle server args (#4217)
This commit is contained in:
@@ -284,9 +284,13 @@ class ServerArgs:
|
|||||||
"Overlap scheduler are disabled because of using "
|
"Overlap scheduler are disabled because of using "
|
||||||
"eagle speculative decoding."
|
"eagle speculative decoding."
|
||||||
)
|
)
|
||||||
# The token generated from the verify step is counted.
|
# The token generated from the verify step is counted in speculative_num_draft_tokens.
|
||||||
# If sepculative_num_steps >= speculative_num_draft_tokens, the additional tokens will definitely be discarded.
|
# If sepculative_num_steps >= speculative_num_draft_tokens, the additional tokens will definitely be discarded.
|
||||||
# assert self.speculative_num_steps < self.speculative_num_draft_tokens
|
assert self.speculative_num_steps < self.speculative_num_draft_tokens
|
||||||
|
assert (
|
||||||
|
self.speculative_num_draft_tokens - 1
|
||||||
|
<= self.speculative_num_steps * self.speculative_eagle_topk
|
||||||
|
)
|
||||||
|
|
||||||
# GGUF
|
# GGUF
|
||||||
if (
|
if (
|
||||||
|
|||||||
Reference in New Issue
Block a user