riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-03 12:50:06 +06:00
Code Issues Actions 28 Packages Projects Releases Wiki Activity

Fix masking utils (#38783)

Browse Source 
* fix

* Update masking_utils.py

* Update masking_utils.py
This commit is contained in:
Cyril Vallez 2025-06-12 11:00:46 +02:00 committed by GitHub
parent 7c58336949
commit 887054c714
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
src/transformers/masking_utils.py
View File

@ -684,8 +684,8 @@ def create_causal_mask(
useful to easily overlay another mask on top of the causal one, for example for image tokens handling. useful to easily overlay another mask on top of the causal one, for example for image tokens handling.
""" """
# If we have an HybridCache structure, here we want to create the mask for the full layers # If we have an HybridCache structure, here we want to create the mask for the full layers
if past_key_values is not None and hasattr(past_key_values, "is_sliding"): if hasattr(past_key_values, "is_sliding") and False in past_key_values.is_sliding:
layer_idx = past_key_values.is_sliding.index(False) if False in past_key_values.is_sliding else 0 layer_idx = past_key_values.is_sliding.index(False)
else: else:
layer_idx = 0 layer_idx = 0
@ -765,9 +765,9 @@ def create_sliding_window_causal_mask(
An optional mask function to combine with the sliding causal mask function (by doing the intersection of both). This is An optional mask function to combine with the sliding causal mask function (by doing the intersection of both). This is
useful to easily overlay another mask on top of the sliding causal one, for example for image tokens handling. useful to easily overlay another mask on top of the sliding causal one, for example for image tokens handling.
""" """
# If we have an HybridCache structure, here we want to create the mask for the full layers # If we have an HybridCache structure, here we want to create the mask for the sliding layers
if past_key_values is not None and hasattr(past_key_values, "is_sliding"): if hasattr(past_key_values, "is_sliding") and True in past_key_values.is_sliding:
layer_idx = past_key_values.is_sliding.index(False) if False in past_key_values.is_sliding else 0 layer_idx = past_key_values.is_sliding.index(True)
else: else:
layer_idx = 0 layer_idx = 0
@ -852,9 +852,9 @@ def create_chunked_causal_mask(
An optional mask function to combine with the chunked causal mask function (by doing the intersection of both). This is An optional mask function to combine with the chunked causal mask function (by doing the intersection of both). This is
useful to easily overlay another mask on top of the chunked causal one, for example for image tokens handling. useful to easily overlay another mask on top of the chunked causal one, for example for image tokens handling.
""" """
# If we have an HybridCache structure, here we want to create the mask for the full layers # If we have an HybridCache structure, here we want to create the mask for the sliding layers
if past_key_values is not None and hasattr(past_key_values, "is_sliding"): if hasattr(past_key_values, "is_sliding") and True in past_key_values.is_sliding:
layer_idx = past_key_values.is_sliding.index(False) if False in past_key_values.is_sliding else 0 layer_idx = past_key_values.is_sliding.index(True)
else: else:
layer_idx = 0 layer_idx = 0