Support penalty in overlap mode; return logprob with chunked prefill; improve benchmark scripts (#3988)
Co-authored-by: SangBin Cho <rkooo567@gmail.com> Co-authored-by: dhou-xai <dhou@x.ai> Co-authored-by: Hanming Lu <hanming_lu@berkeley.edu>
This commit is contained in:
@@ -457,7 +457,7 @@ class VocabParallelEmbedding(torch.nn.Module):
|
||||
assert loaded_weight.shape[output_dim] == (
|
||||
self.org_vocab_size
|
||||
// (self.tp_size if self.use_presharded_weights else 1)
|
||||
)
|
||||
), f"{self.org_vocab_size=} {self.use_presharded_weights=} {loaded_weight.shape[output_dim]=}"
|
||||
|
||||
# Copy the data.
|
||||
if not self.use_presharded_weights:
|
||||
|
||||
Reference in New Issue
Block a user