Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix crash making embeddings from too-long prompts with attention weights

Browse Source
This commit is contained in:
Damian at mba 2022-10-26 21:25:10 +02:00 committed by Lincoln Stein
parent 0d0481ce75
commit 79689e87ce
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
ldm/modules/encoders/modules.py
View File

@ -598,7 +598,7 @@ class WeightedFrozenCLIPEmbedder(FrozenCLIPEmbedder):
fragments,
truncation=True,
max_length=self.max_length,
return_overflowing_tokens=False,
return_overflowing_tokens=True,
padding='do_not_pad',
return_tensors=None, # just give me a list of ints
)['input_ids']
@ -616,8 +616,9 @@ class WeightedFrozenCLIPEmbedder(FrozenCLIPEmbedder):
if (len(all_tokens) + 2) > self.max_length:
excess_token_count = (len(all_tokens) + 2) - self.max_length
print(f"prompt is {excess_token_count} token(s) too long and has been truncated")
print(f">> Prompt is {excess_token_count} token(s) too long and has been truncated")
all_tokens = all_tokens[:self.max_length - 2]
per_token_weights = per_token_weights[:self.max_length - 2]
# pad out to a 77-entry array: [eos_token, <prompt tokens>, eos_token, ..., eos_token]
# (77 = self.max_length)