liger-kernel-nightly 0.5.5.dev20250315175408__py3-none-any.whl → 0.5.5.dev20250316145754__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.


This version of liger-kernel-nightly might be problematic. Click here for more details.

@@ -172,7 +172,7 @@ def lce_forward(
172
172
  shift_labels = shift_labels.contiguous()
173
173
 
174
174
  # Flatten hidden state
175
- shift_hidden_states = shift_hidden_states.view(-1, self.config.hidden_size)
175
+ shift_hidden_states = shift_hidden_states.view(-1, self.config.text_config.hidden_size)
176
176
  shift_labels = shift_labels.view(-1).to(hidden_device)
177
177
 
178
178
  lce = LigerFusedLinearCrossEntropyLoss()
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: liger_kernel_nightly
3
- Version: 0.5.5.dev20250315175408
3
+ Version: 0.5.5.dev20250316145754
4
4
  Summary: Efficient Triton kernels for LLM Training
5
5
  License: BSD 2-CLAUSE LICENSE
6
6
  Copyright 2024 LinkedIn Corporation
@@ -59,7 +59,7 @@ liger_kernel/transformers/model/mistral.py,sha256=MVRksI5_j_8WJu8znOHKCdSI5jSu-S
59
59
  liger_kernel/transformers/model/mixtral.py,sha256=jpZJkpl625Q-JHWarj2MqT5mRaSsiCtg0c9vVyvOdCY,11430
60
60
  liger_kernel/transformers/model/mllama.py,sha256=qWexBdskuN3gPJvPUwt4J0nU675tGD6W7wxgRZ9Bifg,11145
61
61
  liger_kernel/transformers/model/olmo2.py,sha256=yyksS6E4fuWd8asEW8rEDBKqZpFmP4ITCM_bjIDZaoY,5124
62
- liger_kernel/transformers/model/paligemma.py,sha256=29NiiTB2pihsq9OogX4Oium94WT52PBu9F4c5vrAV3s,10286
62
+ liger_kernel/transformers/model/paligemma.py,sha256=C_Pb1qqxZl0J0fyXlwp1jTwNXckK9xuoSLHXy3rkWsE,10298
63
63
  liger_kernel/transformers/model/phi3.py,sha256=biRa8fph9qdnQmkD9I21t5XIjpIt1i6UKU4uk8Up8pU,10292
64
64
  liger_kernel/transformers/model/qwen2.py,sha256=14UuPjxB-tjqWn85Tn4fqBFvVhVsth5iPEt8kJSMiew,9581
65
65
  liger_kernel/transformers/model/qwen2_5_vl.py,sha256=l71WBfX0ptrisoURIRwXJH7MQ2vGKOvcRYMNsrydwlQ,9455
@@ -68,9 +68,9 @@ liger_kernel/transformers/trainer/__init__.py,sha256=p7yQfklV8-467qSz_ZMimkbDF7H
68
68
  liger_kernel/transformers/trainer/orpo_trainer.py,sha256=pdekW7l6Qg_aqa5SYKYlSWUF8m3lkOFvFLcIMEHrz9s,8338
69
69
  liger_kernel/triton/__init__.py,sha256=qCiCamzCRv6lpV8IqpAc9YMdNKC7GKurClWceQPnlis,92
70
70
  liger_kernel/triton/monkey_patch.py,sha256=Rd0hUHAzDkFfHvnX7-PBaNK5EKnZhtfM_h-fgQH9HPY,1568
71
- liger_kernel_nightly-0.5.5.dev20250315175408.dist-info/LICENSE,sha256=OhzLDHJ0to4a8sodVLELZiCFylZ1NAAYLs-HrjPy0ag,1312
72
- liger_kernel_nightly-0.5.5.dev20250315175408.dist-info/METADATA,sha256=QARM55FStQf_GHgPq-tmKrohwU7rCNQ6dPdObQ4q3nU,22588
73
- liger_kernel_nightly-0.5.5.dev20250315175408.dist-info/NOTICE,sha256=njwnoPZLh9AN8SJQzxvCGLHi-8X__AvWRze6joNXIY8,2066
74
- liger_kernel_nightly-0.5.5.dev20250315175408.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
75
- liger_kernel_nightly-0.5.5.dev20250315175408.dist-info/top_level.txt,sha256=2eghu4hA3LnkM7ElW92tQ8zegWKgSbeo-k-aGe1YnvY,13
76
- liger_kernel_nightly-0.5.5.dev20250315175408.dist-info/RECORD,,
71
+ liger_kernel_nightly-0.5.5.dev20250316145754.dist-info/LICENSE,sha256=OhzLDHJ0to4a8sodVLELZiCFylZ1NAAYLs-HrjPy0ag,1312
72
+ liger_kernel_nightly-0.5.5.dev20250316145754.dist-info/METADATA,sha256=sLjW-HmLsWS53YnGPerhzMPFRxydL0nBuzUNnm6uCXk,22588
73
+ liger_kernel_nightly-0.5.5.dev20250316145754.dist-info/NOTICE,sha256=njwnoPZLh9AN8SJQzxvCGLHi-8X__AvWRze6joNXIY8,2066
74
+ liger_kernel_nightly-0.5.5.dev20250316145754.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
75
+ liger_kernel_nightly-0.5.5.dev20250316145754.dist-info/top_level.txt,sha256=2eghu4hA3LnkM7ElW92tQ8zegWKgSbeo-k-aGe1YnvY,13
76
+ liger_kernel_nightly-0.5.5.dev20250316145754.dist-info/RECORD,,