Daniel Han
a5e37ac1a2
Cmake and ninja move
2025-07-01 00:13:00 -07:00
Daniel Han
3c1c8e5b95
Fix setup.py
2025-07-01 00:05:56 -07:00
Daniel Han
08f5c90428
Update _utils.py
2025-06-30 23:13:33 -07:00
Daniel Han
e29cbd8d83
Remove stale bot
2025-06-30 23:11:57 -07:00
billishyahao
06ca5c273b
[Feature] enable unsloth on amd gpu ( #2520 )
...
* [Feature] enable unsloth on amd gpu
* fix the comment
---------
Co-authored-by: Daniel Han <danielhanchen@gmail.com>
2025-06-30 16:52:05 -07:00
Rishabh
a15e77fc63
Convert torch.bfloat16, torch.float16, etc. to vLLM valid dtypes ( #2811 )
...
* Convert torch.bfloat16, torch.float16, etc. to vLLM valid dtypes
* removed newlines and extra whitespace
2025-06-30 16:39:36 -07:00
DoubleMathew
8e6ec3bf04
Fix loftq None config for FastBaseModel ( #2848 )
...
add new validate_loftq_config to __all__
2025-06-30 16:38:51 -07:00
Daniel Han
9aa13e4e64
Gemma 3N bug fixes ( #2842 )
...
* Update vision.py
* Bug fix
* Update mapper.py
* check SDPA for Mistral 3, Pixtral
* Update vision.py
* Versioning
* Update rl_replacements.py
* Update README.md
* add model registry
* move hf hub utils to unsloth/utils
* refactor global model info dicts to dataclasses
* fix dataclass init
* fix llama registration
* remove deprecated key function
* start registry reog
* add llama vision
* quant types -> Enum
* remap literal quant types to QuantType Enum
* add llama model registration
* fix quant tag mapping
* add qwen2.5 models to registry
* add option to include original model in registry
* handle quant types per model size
* separate registration of base and instruct llama3.2
* add QwenQVQ to registry
* add gemma3 to registry
* add phi
* add deepseek v3
* add deepseek r1 base
* add deepseek r1 zero
* add deepseek distill llama
* add deepseek distill models
* remove redundant code when constructing model names
* add mistral small to registry
* rename model registration methods
* rename deepseek registration methods
* refactor naming for mistral and phi
* add global register models
* refactor model registration tests for new registry apis
* add model search method
* remove deprecated registration api
* add quant type test
* add registry readme
* make llama registration more specific
* clear registry when executing individual model registration file
* more registry readme updates
* Update _auto_install.py
* Llama4
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Synthetic data
* Update mapper.py
* Xet and Synthetic
* Update synthetic.py
* Update loader.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update pyproject.toml
* Delete .gitignore
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update _utils.py
* Update pyproject.toml
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update chat_templates.py
* Seasame force float16 / float32
* Fix Seasame
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* is_multimodal
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* UNSLOTH_DISABLE_STATIC_GENERATION
* Update vision.py
* Auto vision detection
* Sesame
* Whisper
* Update loader.py
* Update loader.py
* Update loader.py
* Update mapper.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update _utils.py
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* logging
* Update pyproject.toml
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* logits / temperature
* Update rl_replacements.py
* Update pyproject.toml
* Update rl_replacements.py
* Update rl_replacements.py
* Debugging only
* Update llama.py
* Update llama.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Generic efficient GRPO
* Update rl_replacements.py
* Update rl_replacements.py
* Remove debugging
* Update rl_replacements.py
* Update rl_replacements.py
* Update vision.py
* Update llama.py
* Update rl_replacements.py
* versioning
* Update _utils.py
* Update vision.py
* Update mapper.py
* Update loader.py
* Update mapper.py
* Update vision.py
* Update loader.py
* Update vision.py
* Update loader.py
* Update _utils.py
* Update vision.py
* gradient checkpointing
* Gemma 3N fixes
* Update loader.py
* Versioning
* Gemma 3N fixes
* Update vision.py
* Update vision.py
* Update loader.py
* Update vision.py
---------
Co-authored-by: Jack Shi Wei Lun <87535974+jackswl@users.noreply.github.com>
Co-authored-by: jeromeku <jerome.ku@gmail.com>
Co-authored-by: Michael Han <107991372+shimmyshimmer@users.noreply.github.com>
2025-06-30 07:15:48 -07:00
Roland Tannous
3d78d44e76
Added conda/mamba section to blackwell installation readme ( #2817 )
...
* Added conda/mamba section to blackwell installation readme
* fix conda creation suffix and vllm install syntax
2025-06-30 06:02:57 -07:00
Daniel Han
bc77032d90
Update stale.yml
2025-06-30 02:16:09 -07:00
Daniel Han
651c8d4498
Create stale.yml ( #2836 )
2025-06-29 21:59:43 -07:00
Daniel Han
748b192177
Delete stale.yml
2025-06-29 21:58:55 -07:00
Daniel Han
3e3fdecaf1
Update stale.yml
2025-06-29 21:57:30 -07:00
Daniel Han
19abdf89e7
Create stale.yml ( #2832 )
2025-06-29 17:36:23 -07:00
Daniel Han
dd764704de
Delete stale.yml
2025-06-29 17:35:01 -07:00
Daniel Han
33203d1bac
Update stale.yml
2025-06-29 17:29:36 -07:00
Daniel Han
a63a031547
Create stale.yml
2025-06-29 17:28:18 -07:00
Mehmet Oguz Derin
58a808f171
Fix LoftQ with FastBaseModel ( #2826 )
...
Pass `init_lora_weights` and `loftq_config` to `LoraConfig` constructor, which enables classes like `FastModel` to use LoftQ support. Thank you very much in advance!
2025-06-29 16:14:02 -07:00
DoubleMathew
5d460ba39e
import undefined transformers_version for falcon model ( #2822 )
...
* import undefined transformers_version for falcon model
fixed falcon transformers version check and added error handling for FalconH1Attention bad import
* Also, conditionally load module from falcon_h1 depending on if the transformers version supports is
2025-06-28 17:41:19 -07:00
DoubleMathew
6e72d688f5
granite force layernorm upcast ( #2799 )
2025-06-28 05:27:56 -07:00
Dhia Eddine Rhaiem
a8f3d69266
Add falcon h1 ( #2650 )
...
* add falcon h1
* feat: add Falcon-H1 into unsloth
* address comments
* fix
* Update unsloth/models/llama.py
Co-authored-by: Daniel Han <danielhanchen@gmail.com>
* Update unsloth/models/llama.py
Co-authored-by: Daniel Han <danielhanchen@gmail.com>
* fixes
* fix comments
---------
Co-authored-by: Daniel Han <danielhanchen@gmail.com>
Co-authored-by: Younes B <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Younes Belkada <younes.belkada@tii.ae>
Co-authored-by: ilyasch2 <ilyas.chahed@tii.ae>
2025-06-28 04:55:13 -07:00
jeromeku
b02be210dc
add instructions for installing on blackwell ( #2812 )
2025-06-27 04:54:55 -07:00
Daniel Han
bfa6a3678e
Update loader.py
2025-06-26 12:03:29 -07:00
Daniel Han
1b94959160
Update _utils.py
2025-06-26 11:31:17 -07:00
Daniel Han
5d933d2b4b
Update loader.py
2025-06-26 11:31:08 -07:00
Daniel Han
93a3cbd783
Update pyproject.toml
2025-06-26 09:15:14 -07:00
Daniel Han
2cdcf61b2a
Gemma 3N ( #2809 )
...
* Add QLoRA Train and Merge16bit Test (#2130 )
* add reference and unsloth lora merging tests
* add test / dataset printing to test scripts
* allow running tests from repo root
* add qlora test readme
* more readme edits
* ruff formatting
* additional readme comments
* forgot to add actual tests
* add apache license
* Update pyproject.toml
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Revert
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Bug fix
* Update mapper.py
* check SDPA for Mistral 3, Pixtral
* Update vision.py
* Versioning
* Update rl_replacements.py
* Update README.md
* add model registry
* move hf hub utils to unsloth/utils
* refactor global model info dicts to dataclasses
* fix dataclass init
* fix llama registration
* remove deprecated key function
* start registry reog
* add llama vision
* quant types -> Enum
* remap literal quant types to QuantType Enum
* add llama model registration
* fix quant tag mapping
* add qwen2.5 models to registry
* add option to include original model in registry
* handle quant types per model size
* separate registration of base and instruct llama3.2
* add QwenQVQ to registry
* add gemma3 to registry
* add phi
* add deepseek v3
* add deepseek r1 base
* add deepseek r1 zero
* add deepseek distill llama
* add deepseek distill models
* remove redundant code when constructing model names
* add mistral small to registry
* rename model registration methods
* rename deepseek registration methods
* refactor naming for mistral and phi
* add global register models
* refactor model registration tests for new registry apis
* add model search method
* remove deprecated registration api
* add quant type test
* add registry readme
* make llama registration more specific
* clear registry when executing individual model registration file
* more registry readme updates
* Update _auto_install.py
* Llama4
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Synthetic data
* Update mapper.py
* Xet and Synthetic
* Update synthetic.py
* Update loader.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update pyproject.toml
* Delete .gitignore
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update _utils.py
* Update pyproject.toml
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update chat_templates.py
* Seasame force float16 / float32
* Fix Seasame
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* is_multimodal
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* UNSLOTH_DISABLE_STATIC_GENERATION
* Update vision.py
* Auto vision detection
* Sesame
* Whisper
* Update loader.py
* Update loader.py
* Update loader.py
* Update mapper.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update _utils.py
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* logging
* Update pyproject.toml
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* logits / temperature
* Update rl_replacements.py
* Update pyproject.toml
* Update rl_replacements.py
* Update rl_replacements.py
* Debugging only
* Update llama.py
* Update llama.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Generic efficient GRPO
* Update rl_replacements.py
* Update rl_replacements.py
* Remove debugging
* Update rl_replacements.py
* Update rl_replacements.py
* Update vision.py
* Update llama.py
* Update rl_replacements.py
* versioning
* Update _utils.py
* Update vision.py
* Update mapper.py
* Update loader.py
* Update mapper.py
* Update vision.py
* Update loader.py
* Update vision.py
* Update loader.py
* Update _utils.py
---------
Co-authored-by: jeromeku <jerome.ku@gmail.com>
Co-authored-by: Jack Shi Wei Lun <87535974+jackswl@users.noreply.github.com>
Co-authored-by: Michael Han <107991372+shimmyshimmer@users.noreply.github.com>
2025-06-26 09:14:28 -07:00
Daniel Han
425ae4d076
Bug fixes ( #2807 )
...
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update vision.py
* HF Transfer
* fix(utils): add missing importlib import to fix NameError (#2134 )
This commit fixes a NameError that occurs when `importlib` is referenced in _utils.py
without being imported, especially when UNSLOTH_USE_MODELSCOPE=1 is enabled.
By adding the missing import statement, the code will no longer throw a NameError.
* Add QLoRA Train and Merge16bit Test (#2130 )
* add reference and unsloth lora merging tests
* add test / dataset printing to test scripts
* allow running tests from repo root
* add qlora test readme
* more readme edits
* ruff formatting
* additional readme comments
* forgot to add actual tests
* add apache license
* Update pyproject.toml
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Revert
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Bug fix
* Update mapper.py
* check SDPA for Mistral 3, Pixtral
* Update vision.py
* Versioning
* Update rl_replacements.py
* Update README.md
* add model registry
* move hf hub utils to unsloth/utils
* refactor global model info dicts to dataclasses
* fix dataclass init
* fix llama registration
* remove deprecated key function
* start registry reog
* add llama vision
* quant types -> Enum
* remap literal quant types to QuantType Enum
* add llama model registration
* fix quant tag mapping
* add qwen2.5 models to registry
* add option to include original model in registry
* handle quant types per model size
* separate registration of base and instruct llama3.2
* add QwenQVQ to registry
* add gemma3 to registry
* add phi
* add deepseek v3
* add deepseek r1 base
* add deepseek r1 zero
* add deepseek distill llama
* add deepseek distill models
* remove redundant code when constructing model names
* add mistral small to registry
* rename model registration methods
* rename deepseek registration methods
* refactor naming for mistral and phi
* add global register models
* refactor model registration tests for new registry apis
* add model search method
* remove deprecated registration api
* add quant type test
* add registry readme
* make llama registration more specific
* clear registry when executing individual model registration file
* more registry readme updates
* Update _auto_install.py
* Llama4
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Synthetic data
* Update mapper.py
* Xet and Synthetic
* Update synthetic.py
* Update loader.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update pyproject.toml
* Delete .gitignore
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update _utils.py
* Update pyproject.toml
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update chat_templates.py
* Seasame force float16 / float32
* Fix Seasame
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* is_multimodal
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* UNSLOTH_DISABLE_STATIC_GENERATION
* Update vision.py
* Auto vision detection
* Sesame
* Whisper
* Update loader.py
* Update loader.py
* Update loader.py
* Update mapper.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update _utils.py
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* logging
* Update pyproject.toml
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* logits / temperature
* Update rl_replacements.py
* Update pyproject.toml
* Update rl_replacements.py
* Update rl_replacements.py
* Debugging only
* Update llama.py
* Update llama.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Generic efficient GRPO
* Update rl_replacements.py
* Update rl_replacements.py
* Remove debugging
* Update rl_replacements.py
* Update rl_replacements.py
* Update vision.py
* Update llama.py
* Update rl_replacements.py
* versioning
* Update _utils.py
* Update vision.py
---------
Co-authored-by: naliazheli <nalia0316@gmail.com>
Co-authored-by: jeromeku <jerome.ku@gmail.com>
Co-authored-by: Jack Shi Wei Lun <87535974+jackswl@users.noreply.github.com>
Co-authored-by: Michael Han <107991372+shimmyshimmer@users.noreply.github.com>
2025-06-26 04:40:47 -07:00
Daniel Han
b371c6b87b
Bug fixes ( #2805 )
...
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update vision.py
* HF Transfer
* fix(utils): add missing importlib import to fix NameError (#2134 )
This commit fixes a NameError that occurs when `importlib` is referenced in _utils.py
without being imported, especially when UNSLOTH_USE_MODELSCOPE=1 is enabled.
By adding the missing import statement, the code will no longer throw a NameError.
* Add QLoRA Train and Merge16bit Test (#2130 )
* add reference and unsloth lora merging tests
* add test / dataset printing to test scripts
* allow running tests from repo root
* add qlora test readme
* more readme edits
* ruff formatting
* additional readme comments
* forgot to add actual tests
* add apache license
* Update pyproject.toml
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Revert
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Bug fix
* Update mapper.py
* check SDPA for Mistral 3, Pixtral
* Update vision.py
* Versioning
* Update rl_replacements.py
* Update README.md
* add model registry
* move hf hub utils to unsloth/utils
* refactor global model info dicts to dataclasses
* fix dataclass init
* fix llama registration
* remove deprecated key function
* start registry reog
* add llama vision
* quant types -> Enum
* remap literal quant types to QuantType Enum
* add llama model registration
* fix quant tag mapping
* add qwen2.5 models to registry
* add option to include original model in registry
* handle quant types per model size
* separate registration of base and instruct llama3.2
* add QwenQVQ to registry
* add gemma3 to registry
* add phi
* add deepseek v3
* add deepseek r1 base
* add deepseek r1 zero
* add deepseek distill llama
* add deepseek distill models
* remove redundant code when constructing model names
* add mistral small to registry
* rename model registration methods
* rename deepseek registration methods
* refactor naming for mistral and phi
* add global register models
* refactor model registration tests for new registry apis
* add model search method
* remove deprecated registration api
* add quant type test
* add registry readme
* make llama registration more specific
* clear registry when executing individual model registration file
* more registry readme updates
* Update _auto_install.py
* Llama4
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Synthetic data
* Update mapper.py
* Xet and Synthetic
* Update synthetic.py
* Update loader.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update pyproject.toml
* Delete .gitignore
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update _utils.py
* Update pyproject.toml
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update chat_templates.py
* Seasame force float16 / float32
* Fix Seasame
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* is_multimodal
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* UNSLOTH_DISABLE_STATIC_GENERATION
* Update vision.py
* Auto vision detection
* Sesame
* Whisper
* Update loader.py
* Update loader.py
* Update loader.py
* Update mapper.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update _utils.py
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* logging
* Update pyproject.toml
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* logits / temperature
* Update rl_replacements.py
* Update pyproject.toml
* Update rl_replacements.py
* Update rl_replacements.py
* Debugging only
* Update llama.py
* Update llama.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl_replacements.py
* Generic efficient GRPO
* Update rl_replacements.py
* Update rl_replacements.py
* Remove debugging
* Update rl_replacements.py
* Update rl_replacements.py
* Update vision.py
* Update llama.py
* Update rl_replacements.py
* versioning
---------
Co-authored-by: naliazheli <nalia0316@gmail.com>
Co-authored-by: jeromeku <jerome.ku@gmail.com>
Co-authored-by: Jack Shi Wei Lun <87535974+jackswl@users.noreply.github.com>
Co-authored-by: Michael Han <107991372+shimmyshimmer@users.noreply.github.com>
2025-06-26 02:17:12 -07:00
Datta Nimmaturi
b81c333b39
Fix grpo sleep regex and indentation ( #2804 )
2025-06-26 01:50:47 -07:00
Lei Zhenyuan
396531ec1f
fix for inductor no attribute prop.multi_processor_count ( #2803 )
2025-06-26 01:44:15 -07:00
DoubleMathew
6e74172c7c
[4/N] Enable intel GPU for unsloth ( #2801 )
...
* add code for xpu llama
* refine code
* change version check to 2.6.0
* remove unuse blank
* reslove commits
* Cleaned up statistics printing
* Update unsloth/models/_utils.py
---------
Co-authored-by: lei,zhenyuan <zhenyuan.lei@intel.com>
Co-authored-by: Daniel Han <danielhanchen@gmail.com>
2025-06-25 20:28:07 -07:00
Michael Han
f14d7e1542
Update README.md
...
Updating links
2025-06-25 01:32:24 -07:00
Lei Zhenyuan
01c5e1a249
[3/N] Enable intel GPU for unsloth ( #2620 )
...
* enable intel xpu changes within kernels
* reslove torch.version < 2.6
* change version check to 2.6.0
* resolve comments for torch_gpu_device
* resolve amp fwd comments
* fix typo
* change cuda default logic
* clean this pr
* add HAS_CUDA_STREAM as default False
* split GPU streams to cuda and xpu streams
* add optional
2025-06-24 02:01:28 -07:00
DoubleMathew
12a05c2293
move min_sms in is_big_gpu inside DEVICE_TYPE if else ( #2792 )
...
log is not defined in torch inductor so remove
Remove log.warning entirely
2025-06-23 18:57:55 -07:00
pluesclues
0211bf63a0
Fixed Sequence Classification errors, loaded model weirdly ( #2793 )
2025-06-23 18:56:56 -07:00
Michael Han
33c304884d
Update issue templates
2025-06-23 05:34:46 -07:00
Daniel Han
e90ea2be0c
Update issue templates
2025-06-23 05:26:28 -07:00
Lei Zhenyuan
d95960b7ce
[5/N] Enable intel GPU for unsloth ( #2768 )
...
* add is_big_gpu support for xpu
* make code unsloth's style
2025-06-23 04:47:34 -07:00
kilavvy
5ed41c1b33
Docs: Fix typo and improve MoE docstrings ( #2784 )
...
* Update qwen3_moe.py
* Update interface.py
2025-06-23 01:09:23 -07:00
Daniel Han
e2aafc7ee9
Fix GRPO ( #2787 )
...
* Update _utils.py
* Update _utils.py
* versioning
* Update _utils.py
* Update _utils.py
* Update _utils.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update llama.py
* Update vision.py
* HF Transfer
* fix(utils): add missing importlib import to fix NameError (#2134 )
This commit fixes a NameError that occurs when `importlib` is referenced in _utils.py
without being imported, especially when UNSLOTH_USE_MODELSCOPE=1 is enabled.
By adding the missing import statement, the code will no longer throw a NameError.
* Add QLoRA Train and Merge16bit Test (#2130 )
* add reference and unsloth lora merging tests
* add test / dataset printing to test scripts
* allow running tests from repo root
* add qlora test readme
* more readme edits
* ruff formatting
* additional readme comments
* forgot to add actual tests
* add apache license
* Update pyproject.toml
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Revert
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Bug fix
* Update mapper.py
* check SDPA for Mistral 3, Pixtral
* Update vision.py
* Versioning
* Update rl_replacements.py
* Update README.md
* add model registry
* move hf hub utils to unsloth/utils
* refactor global model info dicts to dataclasses
* fix dataclass init
* fix llama registration
* remove deprecated key function
* start registry reog
* add llama vision
* quant types -> Enum
* remap literal quant types to QuantType Enum
* add llama model registration
* fix quant tag mapping
* add qwen2.5 models to registry
* add option to include original model in registry
* handle quant types per model size
* separate registration of base and instruct llama3.2
* add QwenQVQ to registry
* add gemma3 to registry
* add phi
* add deepseek v3
* add deepseek r1 base
* add deepseek r1 zero
* add deepseek distill llama
* add deepseek distill models
* remove redundant code when constructing model names
* add mistral small to registry
* rename model registration methods
* rename deepseek registration methods
* refactor naming for mistral and phi
* add global register models
* refactor model registration tests for new registry apis
* add model search method
* remove deprecated registration api
* add quant type test
* add registry readme
* make llama registration more specific
* clear registry when executing individual model registration file
* more registry readme updates
* Update _auto_install.py
* Llama4
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Synthetic data
* Update mapper.py
* Xet and Synthetic
* Update synthetic.py
* Update loader.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update pyproject.toml
* Delete .gitignore
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update _utils.py
* Update pyproject.toml
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update synthetic.py
* Update chat_templates.py
* Seasame force float16 / float32
* Fix Seasame
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* is_multimodal
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update vision.py
* Update vision.py
* Update vision.py
* UNSLOTH_DISABLE_STATIC_GENERATION
* Update vision.py
* Auto vision detection
* Sesame
* Whisper
* Update loader.py
* Update loader.py
* Update loader.py
* Update mapper.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update vision.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update loader.py
* Update _utils.py
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* Update rl.py
* logging
* Update pyproject.toml
* Update rl.py
* versioning
* Update rl.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* Update rl.py
* Update rl_replacements.py
* Update rl_replacements.py
* logits / temperature
* Update rl_replacements.py
* Update pyproject.toml
* Update rl_replacements.py
* Update rl_replacements.py
---------
Co-authored-by: naliazheli <nalia0316@gmail.com>
Co-authored-by: jeromeku <jerome.ku@gmail.com>
Co-authored-by: Jack Shi Wei Lun <87535974+jackswl@users.noreply.github.com>
Co-authored-by: Michael Han <107991372+shimmyshimmer@users.noreply.github.com>
2025-06-22 05:54:29 -07:00
Daniel Han
8767244b6e
Update rl.py
2025-06-21 23:29:08 -07:00
Daniel Han
41421bfb52
Fix bf16 = None
2025-06-21 22:50:29 -07:00
Daniel Han
798530797b
Update rl.py
2025-06-21 22:26:46 -07:00
Daniel Han
e82cc4ba8a
Update _utils.py
2025-06-21 22:20:46 -07:00
Daniel Han
02d7b01b37
Update rl_replacements.py
2025-06-21 22:20:32 -07:00
Daniel Han
d8a29af1cb
Fix DAPO, TRL 0.19.0
2025-06-21 22:14:21 -07:00
simpissa
8242205636
Fix for grpo_compute_loss_slow ( #2702 )
...
* slice last logit
* move slicing
2025-06-21 21:58:06 -07:00
Daniel Han
5b9f59600e
Mistral Small 3.2
2025-06-21 06:44:14 -07:00
amrothemich
9a4ca12b70
Update pyproject.toml ( #2778 )
...
Switched pyproject license to dictionary type
2025-06-21 02:44:24 -07:00
