apunkt/doc-to-lora
1
0
Fork 0
mirror of https://github.com/SakanaAI/doc-to-lora.git synced 2026-07-23 17:01:04 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix attention mask flash attn perceiver

Browse source
This commit is contained in:
51616 2025-01-29 15:25:07 +00:00
parent 6864424c68
commit e7a0569745
1 changed files with 1 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
src/ctx_to_lora/modeling_idefics2.py
View file

@ -663,6 +663,7 @@ class Idefics2PerceiverResampler(Idefics2PreTrainedModel):
seq_lens_k = attention_mask.sum(dim=-1, dtype=torch.int32)
cu_seq_lens_k = torch.cumsum(seq_lens_k, dim=0, dtype=torch.int32)
max_length_k = seq_lens_k.max().item()
cu_seq_lens_k = nn.functional.pad(cu_seq_lens_k, (1, 0))
elif position_ids is not None:
logger.warning_once("Using position ids for resampler")