huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 04:14:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: add missing accepted_ids to batch_top_tokens

Browse Source
This commit is contained in:
drbh 2024-02-07 03:57:35 +00:00
parent 48624fee25
commit 2c6ef7c93a
1 changed files with 3 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
server/text_generation_server/models/mamba.py
View File

@ -489,10 +489,13 @@ class Mamba(Model):
generations: List[Generation] = [] generations: List[Generation] = []
stopped = True stopped = True
# Speculation is not active for causal
accepted_ids = torch.ones_like(batch.input_ids)[:, 0]
batch_top_token_ids, batch_top_token_logprobs = batch_top_tokens( batch_top_token_ids, batch_top_token_logprobs = batch_top_tokens(
batch.top_n_tokens, batch.top_n_tokens,
batch.top_n_tokens_tensor, batch.top_n_tokens_tensor,
torch.log_softmax(logits[:, -1], -1), torch.log_softmax(logits[:, -1], -1),
accepted_ids,
) )
start_decode = time.time_ns() start_decode = time.time_ns()