huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-04-26 20:42:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

Cleaner condition.

Browse Source
This commit is contained in:
Nicolas Patry 2024-09-23 15:50:43 +02:00
parent 8abdd08ef4
commit f1f9079ec6
No known key found for this signature in database
GPG Key ID: 64AF4752B2967863
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
server/text_generation_server/models/idefics_causal_lm.py
View File

@ -225,7 +225,7 @@ class IdeficsCausalLMBatch(Batch):
aspect_ratio_ids = None aspect_ratio_ids = None
aspect_ratio_mask = None aspect_ratio_mask = None
cross_attention_mask = None cross_attention_mask = None
else: elif "cross_attention_mask" in tokenized_inputs:
image_attention_mask = None image_attention_mask = None
aspect_ratio_ids = tokenized_inputs["aspect_ratio_ids"] aspect_ratio_ids = tokenized_inputs["aspect_ratio_ids"]
aspect_ratio_mask = tokenized_inputs["aspect_ratio_mask"] aspect_ratio_mask = tokenized_inputs["aspect_ratio_mask"]
@ -235,6 +235,8 @@ class IdeficsCausalLMBatch(Batch):
tokenized_inputs["input_ids"] = tokenized_inputs["input_ids"].clamp( tokenized_inputs["input_ids"] = tokenized_inputs["input_ids"].clamp(
max=processor.tokenizer.vocab_size - 1 max=processor.tokenizer.vocab_size - 1
) )
else:
raise RuntimeError("Unhandled state for idefics/mllama")
position_ids = tokenized_inputs["attention_mask"].long().cumsum(-1) - 1 position_ids = tokenized_inputs["attention_mask"].long().cumsum(-1) - 1
position_ids.masked_fill_(tokenized_inputs["attention_mask"] == 0, 1) position_ids.masked_fill_(tokenized_inputs["attention_mask"] == 0, 1)