liger-kernel-nightly 0.5.2.dev20241212060959__py3-none-any.whl → 0.5.2.dev20241212071131__py3-none-any.whl
Sign up to get free protection for your applications and to get access to all the features.
- {liger_kernel_nightly-0.5.2.dev20241212060959.dist-info → liger_kernel_nightly-0.5.2.dev20241212071131.dist-info}/METADATA +2 -2
- {liger_kernel_nightly-0.5.2.dev20241212060959.dist-info → liger_kernel_nightly-0.5.2.dev20241212071131.dist-info}/RECORD +6 -6
- {liger_kernel_nightly-0.5.2.dev20241212060959.dist-info → liger_kernel_nightly-0.5.2.dev20241212071131.dist-info}/LICENSE +0 -0
- {liger_kernel_nightly-0.5.2.dev20241212060959.dist-info → liger_kernel_nightly-0.5.2.dev20241212071131.dist-info}/NOTICE +0 -0
- {liger_kernel_nightly-0.5.2.dev20241212060959.dist-info → liger_kernel_nightly-0.5.2.dev20241212071131.dist-info}/WHEEL +0 -0
- {liger_kernel_nightly-0.5.2.dev20241212060959.dist-info → liger_kernel_nightly-0.5.2.dev20241212071131.dist-info}/top_level.txt +0 -0
@@ -1,6 +1,6 @@
|
|
1
1
|
Metadata-Version: 2.1
|
2
2
|
Name: liger_kernel_nightly
|
3
|
-
Version: 0.5.2.
|
3
|
+
Version: 0.5.2.dev20241212071131
|
4
4
|
Summary: Efficient Triton kernels for LLM Training
|
5
5
|
License: BSD 2-CLAUSE LICENSE
|
6
6
|
Copyright 2024 LinkedIn Corporation
|
@@ -136,7 +136,7 @@ With one line of code, Liger Kernel can increase throughput by more than 20% and
|
|
136
136
|
> - Benchmark conditions: LLaMA 3-8B, Batch Size = 8, Data Type = `bf16`, Optimizer = AdamW, Gradient Checkpointing = True, Distributed Strategy = FSDP1 on 8 A100s.
|
137
137
|
> - Hugging Face models start to OOM at a 4K context length, whereas Hugging Face + Liger Kernel scales up to 16K.
|
138
138
|
|
139
|
-
## Optimize
|
139
|
+
## Optimize Post Training with Liger Kernel
|
140
140
|
|
141
141
|
<p align="center">
|
142
142
|
<img src="https://raw.githubusercontent.com/linkedin/Liger-Kernel/main/docs/images/post-training.png" width="50%" alt="Post Training">
|
@@ -58,9 +58,9 @@ liger_kernel/transformers/trainer/__init__.py,sha256=c4OQVJmhNOloj0JYSEc0j_cQuBb
|
|
58
58
|
liger_kernel/transformers/trainer/orpo_trainer.py,sha256=jko6oq_XQdBSmXubp05E-_YXOyhtB5Bj75dg5YNwOsE,7517
|
59
59
|
liger_kernel/triton/__init__.py,sha256=yfRe0zMb47QnqjecZWG7LnanfCTzeku7SgWRAwNVmzU,101
|
60
60
|
liger_kernel/triton/monkey_patch.py,sha256=5BcGKTtdqeYchypBIBopGIWPx1-cFALz7sOKoEsqXJ0,1584
|
61
|
-
liger_kernel_nightly-0.5.2.
|
62
|
-
liger_kernel_nightly-0.5.2.
|
63
|
-
liger_kernel_nightly-0.5.2.
|
64
|
-
liger_kernel_nightly-0.5.2.
|
65
|
-
liger_kernel_nightly-0.5.2.
|
66
|
-
liger_kernel_nightly-0.5.2.
|
61
|
+
liger_kernel_nightly-0.5.2.dev20241212071131.dist-info/LICENSE,sha256=OhzLDHJ0to4a8sodVLELZiCFylZ1NAAYLs-HrjPy0ag,1312
|
62
|
+
liger_kernel_nightly-0.5.2.dev20241212071131.dist-info/METADATA,sha256=OXd0vITMpiCdQW9JuXmKsjTwg612utOa0p9biJBsEgo,21055
|
63
|
+
liger_kernel_nightly-0.5.2.dev20241212071131.dist-info/NOTICE,sha256=njwnoPZLh9AN8SJQzxvCGLHi-8X__AvWRze6joNXIY8,2066
|
64
|
+
liger_kernel_nightly-0.5.2.dev20241212071131.dist-info/WHEEL,sha256=P9jw-gEje8ByB7_hXoICnHtVCrEwMQh-630tKvQWehc,91
|
65
|
+
liger_kernel_nightly-0.5.2.dev20241212071131.dist-info/top_level.txt,sha256=2eghu4hA3LnkM7ElW92tQ8zegWKgSbeo-k-aGe1YnvY,13
|
66
|
+
liger_kernel_nightly-0.5.2.dev20241212071131.dist-info/RECORD,,
|
File without changes
|
File without changes
|
File without changes
|