Daniel Han
efd07cd9a7
Auto vision detection
2025-05-15 06:39:03 -07:00
Daniel Han
d4975ca143
Update vision.py
2025-05-15 04:52:10 -07:00
Daniel Han
2b0fb84793
UNSLOTH_DISABLE_STATIC_GENERATION
2025-05-15 04:48:42 -07:00
Daniel Han
1d2ac1f1e8
Update vision.py
2025-05-15 04:34:49 -07:00
Daniel Han
a8884bc8c1
Update vision.py
2025-05-15 04:26:05 -07:00
Daniel Han
e99dfc2589
Update vision.py
2025-05-15 03:54:22 -07:00
Daniel Han
98e48b88c0
Update loader.py
2025-05-15 03:46:48 -07:00
Daniel Han
6f00d6d3a9
Update loader.py
2025-05-15 03:39:20 -07:00
Daniel Han
6a1904c704
Update loader.py
2025-05-15 03:18:16 -07:00
Daniel Han
0eba0ed5b3
Update loader.py
2025-05-15 03:00:25 -07:00
Daniel Han
4f8027ad74
is_multimodal
2025-05-15 02:32:29 -07:00
Daniel Han
bcdd2d52e5
Update loader.py
2025-05-15 02:30:12 -07:00
Daniel Han
b4d0680de4
Update vision.py
2025-05-15 02:17:03 -07:00
Daniel Han
924a6f4952
Update vision.py
2025-05-15 02:07:29 -07:00
Daniel Han
ab31b9af62
Update vision.py
2025-05-15 01:54:05 -07:00
Daniel Han
505c27cd4c
Update loader.py
2025-05-15 01:47:34 -07:00
Daniel Han
3a9c10c33c
Fix Seasame
2025-05-15 01:45:15 -07:00
Daniel Han
8f42a99e49
Seasame force float16 / float32
2025-05-15 01:27:14 -07:00
Daniel Han
e51bec7618
Update chat_templates.py
2025-05-15 01:13:20 -07:00
Daniel Han
44053f4cfb
Merge branch 'main' into nightly
2025-05-15 01:07:32 -07:00
Michael Han
e01f304c6d
Merge pull request #2537 from kiankyars/main
...
Add Qwen-3 chat template and Ollama template support
2025-05-14 20:45:53 -07:00
Daniel Han
49c4bf1471
Merge branch 'main' into nightly
2025-05-14 20:30:41 -07:00
Kian Kyars
4e7b91f11d
undo accident
2025-05-14 19:00:23 -06:00
Kian Kyars
973a0ddcd4
style: Place Qwen-3 template after Gemma-3, match style with other templates
2025-05-14 18:59:17 -06:00
Kian Kyars
d3fdd4193b
Update Qwen-3 chat and Ollama templates to official full version, placed after Gemma-3
2025-05-14 18:42:55 -06:00
Kian Kyars
31887228d6
Add Qwen-3 chat template and Ollama template support
2025-05-14 18:35:02 -06:00
Daniel Han
4170900b47
Update pyproject.toml
2025-05-14 05:42:47 -07:00
Daniel Han
75cdebf9b7
Update _utils.py
2025-05-14 05:42:11 -07:00
Daniel Han
957c203f0e
Update synthetic.py
2025-05-14 04:05:23 -07:00
Daniel Han
62e669b291
Update synthetic.py
2025-05-14 03:54:46 -07:00
Daniel Han
9dd46b5a6d
Update synthetic.py
2025-05-14 03:49:27 -07:00
Daniel Han
175c39e9b1
Update synthetic.py
2025-05-14 03:47:29 -07:00
Daniel Han
d6fbd3043e
Update synthetic.py
2025-05-14 03:44:15 -07:00
Michael Han
7d608fef58
Merge pull request #2527 from mmathew23/csm
...
Add Sesame CSM
2025-05-14 02:26:12 -07:00
DoubleMathew
89448d22a7
Merge branch 'unslothai:main' into csm
2025-05-13 17:58:11 -05:00
Daniel Han
77ebc7b74e
Versioning
2025-05-13 09:10:24 -07:00
Daniel Han
71b22abd54
Update synthetic.py
2025-05-13 08:26:17 -07:00
Daniel Han
87adce5e61
Update synthetic.py
2025-05-13 08:25:48 -07:00
Daniel Han
829b79c050
Merge branch 'main' into nightly
2025-05-13 03:39:49 -07:00
Michael Han
99e12bcc78
Update README.md
2025-05-13 01:39:59 -07:00
Daniel Han
1b71e6d542
Update loader_utils.py
2025-05-12 21:06:30 -07:00
Daniel Han
c0ac1cdb01
Update pyproject.toml
2025-05-12 16:28:50 -07:00
feng lui
c9ca2096b7
vLLM Windows CUDA support [tested] ( #2158 )
...
* Update loader.py
change vllm installed check by transformers utils function
* Update llama.py
change vllm installed check by transformers utils function
* add sample notebook
* fix Indentation
* add global is_vLLM_available function
* Pythonic style
* Delete nb/Qwen2.5_(3B)-GRPO-windows.ipynb
Would be great to move it to https://github.com/unslothai/notebooks - appreciate it!
---------
Co-authored-by: Daniel Han <danielhanchen@gmail.com>
2025-05-12 05:33:42 -07:00
Daniel Han
32ba81f076
Update pyproject.toml
2025-05-12 04:24:18 -07:00
Daniel Han
ac5fc0bc46
Fix Intel GPU
2025-05-12 03:10:21 -07:00
Lei Zhenyuan
6d61258dd3
[2/N] Enable intel GPU for unsloth ( #2388 )
...
* add DEVICE_TYPE and resolve device specific API
* reuse import torch
* move env under device typr
* resolve comments
* add more comments
* add more comments
2025-05-12 02:58:21 -07:00
Lei Zhenyuan
7215651d84
first pr for intel GPU, resolve __init__.py and pyproject.toml ( #2350 )
...
add better comments
2025-05-12 02:38:40 -07:00
Daniel Han
47c9ce4ab3
Fix GRPO eval
2025-05-12 02:35:31 -07:00
Michael Han
05bb37893a
Merge pull request #2466 from mmathew23/fix_pop_token_type_ids
...
the pixtral vision notebook fails during inference
2025-05-09 14:57:31 -07:00
Mathew Mathew
80daf6bb83
turn off compilation and fast generation for csm
2025-05-09 16:50:16 -05:00
