ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
4,859 commits 391 branches 58 tags 304 MiB
Commit graph

4,859 commits

This branch
This branch
All branches
Author SHA1 Message Date
Roland Tannous
811a9243b2 Merge pull request #179 from unslothai/fix/copy-mac 
Added the copy feature on mac
 2026-02-20 14:38:29 +04:00
Shine1i
d378e48c2a feat: introduce execution tracking and analysis for recipe preview 
- Added `ExecutionsView` with execution history tracking, live updates, and detailed data analysis.
- Implemented IndexedDB support via Dexie to persist execution records locally.
- Enhanced backend preview logic to return execution analysis and artifacts.
- Updated studio header with view toggling between "Editor" and "Executions."
 2026-02-20 11:34:25 +01:00
Roland Tannous
cdd1f7fce2 Merge pull request #166 from unslothai/feat/download-progress-indicator 
feat: add download progress indicators for dataset preview and training overlay
 2026-02-20 14:11:07 +04:00
Roland Tannous
168055a267 Merge pull request #172 from unslothai/fix/training-param 
Added optim and lr_scheduler_type in the frontend
 2026-02-20 14:05:33 +04:00
Shine1i
763001b78e refactor: remove Jinja autocomplete components and simplify variable handling 
- Deleted `jinja-ref-autocomplete` components and related hooks.
- Replaced custom Jinja variable autocomplete with standard `Textarea` and `Input` components.
- Streamlined variable handling logic by replacing `getAvailableRefItems` with `getAvailableVariables`.
- Removed unused state (`flowMoving`) and redundant logic tied to Jinja-specific functionality.
 2026-02-20 10:36:51 +01:00
samit
3d403c6c99 edited the font of the new parameters 2026-02-20 01:29:37 -08:00
Wasim Yousef Said
6dd0e11439 Merge branch 'nightly' into feature/canvas-lab 2026-02-20 01:23:44 -08:00
Roland Tannous
f61392cfb3 Merge pull request #173 from unslothai/fix/hf-token-model-hiding 
added hf token validation
 2026-02-20 13:19:31 +04:00
samit
5ba8edf9fe added the copy on mac 2026-02-20 00:49:38 -08:00
Roland Tannous
a8c1f5fa84 added NODE OPTIONS export , updating npm to 2.2.6 2026-02-20 08:27:09 +00:00
samit
68028bf7f3 added lr_scheduler type to the frontend 2026-02-19 23:31:46 -08:00
Roland Tannous
fa555c72e8 Merge pull request #175 from unslothai/fix/changing-num-proc-for-filtering 
Fix/changing num proc for filtering
 2026-02-20 10:39:11 +04:00
Roland Tannous
3c1eff525f Merge pull request #171 from unslothai/fix/compare-mode-race-and-adapter-toggle 
Fixing compare feature
 2026-02-20 10:38:56 +04:00
Manan17
798bfb8f6f Setting it to total cpu_count // 4 2026-02-20 06:32:01 +00:00
samit
035f765130 added hf token validation 2026-02-19 21:05:27 -08:00
samit
93b31f0db2 added optim in the frontend 2026-02-19 15:29:23 -08:00
Manan17
fdeccec259 Fixing compare feature 2026-02-19 20:15:44 +00:00
Roland Tannous
b6799a21a5 Merge pull request #169 from unslothai/feature/backwards-compatibility-unsloth-ui-command 
Add `unsloth-ui` alias for backwards compatibility
 2026-02-19 22:44:41 +04:00
Roland Tannous
353cb13618 add unsloth-ui shell alias for backwards compatibility alongside unsloth-studio 2026-02-19 18:41:10 +00:00
Roland Tannous
53820f0eed Merge pull request #168 from unslothai/fix/cli-studio-command-update 
Update `studio` CLI command to use new FastAPI backend
 2026-02-19 22:17:37 +04:00
Roland Tannous
d486a2ef6f rename cli studio command to use new FastAPI backend and add unsloth-ui alias for backwards compatibility 2026-02-19 17:59:22 +00:00
Daniel Han
3bddfed117 Patch trunc_normal_ for low-precision stability (#4027) 
* Fix low-precision trunc_normal initialization instability

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Document TorchTitan trunc_normal low-precision failure mode

* Fix trunc_normal generator positional compatibility

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Fix trunc_normal generator TypeError fallback

---------

Co-authored-by: Daniel Hanchen <danielhanchen@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
 2026-02-19 04:40:14 -08:00
Daniel van Strien
8165266a37 Add optional datasets metadata support to save/push functions (#4076) 
* Add `datasets` metadata support to model cards

Add an optional `datasets` parameter to all save/push functions so users
can specify which datasets were used for training. The metadata is set
via `ModelCard.data.datasets` for standard paths and via
`metadata_update` for GGUF and generic save paths.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Fix datasets metadata for existing repos, add token, improve errors

- Add metadata_update fallback in create_huggingface_repo and
  upload_to_huggingface so datasets metadata is set even when the
  repo already exists (previously only worked on first creation).
- Pass token=token to all metadata_update calls so they work
  without a global HF login.
- Replace silent except:pass with logger.warning_once for
  metadata failures so users know if something went wrong.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Fix generic datasets metadata repo resolution for PR #4076

* Fix create_huggingface_repo username resolution for PR #4076

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Daniel Hanchen <danielhanchen@users.noreply.github.com>
 2026-02-19 03:53:35 -08:00
Roland Tannous
f106ca5ed4 Merge pull request #167 from unslothai/fix/train_on_completions-dataset-check-improvements 
Simplify dataset check to 2-tier, improve multimodal detection, auto-set trainOnCompletions, recheck dataset on reload
 2026-02-19 15:35:50 +04:00
Roland Tannous
18c41c2b08 Simplify dataset check to 2-tier, improve multimodal detection, auto-set trainOnCompletions, recheck dataset on reload 2026-02-19 11:25:54 +00:00
imagineer99
ee6d33fa32 feat: add download progress indicators for dataset preview and training overlay 2026-02-19 08:22:23 +00:00
Roland Tannous
21ee32df40 Auto-set trainOnCompletions based on vision/multimodal state, default all model configs to true, and re-fetch model defaults on page reload 2026-02-19 06:40:40 +00:00
Roland Tannous
e2b7b4b54c change train_on_completions to true 2026-02-19 06:02:16 +00:00
Roland Tannous
9f1caa1ceb Revert "Setting default for the train on responses only" 
This reverts commit 982a855df7.
 2026-02-19 09:55:21 +04:00
Roland Tannous
b35798c8ff Merge pull request #163 from unslothai/fix/add-vision-and-dataset-to-local 
Fix/add vision and dataset to local
 2026-02-19 09:46:54 +04:00
Roland Tannous
35194e4382 Merge pull request #161 from unslothai/fix/vision-model-detection 
Fixing Vision model detection
 2026-02-19 09:43:30 +04:00
Manan17
b7b4cbc949 Adding vision and multimodal dataset to the localstorage 2026-02-19 05:06:41 +00:00
Manan17
982a855df7 Setting default for the train on responses only 2026-02-19 02:59:48 +00:00
Manan17
56869c63bd Passing use_auth = True and also having different checks which is missed by the is_vision function 2026-02-19 02:55:46 +00:00
Kaitao Yang
fd38dc96c3 reduce code duplicaton by inheritting from LlamaRotaryEmbedding (#3878) 
* simplify_code_using_apply_time_scaling

* modify LlamaRotaryEmbedding for better inheritance

* reduce_code_duplication_LlamaExtendedRotaryEmbedding
 2026-02-18 19:13:33 -06:00
Roland Tannous
559ba976bc Merge pull request #159 from unslothai/fix/reduce_dataset_num_proc_25_pct 
Fix/reduce dataset num proc 25 pct
 2026-02-19 00:55:02 +04:00
Roland Tannous
adc0c78dbc reduce dataset_num_proc to 1/4 of cpu_count 2026-02-18 20:53:32 +00:00
Roland Tannous
6aceaec323 Merge pull request #155 from unslothai/fix/sft-tokenizer-unwrap-for-vlm-text 
fix: Unwrap ProcessorMixin to raw tokenizer for text-only SFTTrainer on VLM-architecture models
 2026-02-18 23:19:57 +04:00
Roland Tannous
a5529fbb0e fix: unwrap ProcessorMixin to raw tokenizer for text-only SFTTrainer on VLM-architecture models 2026-02-18 19:16:45 +00:00
Roland Tannous
e33920974b fix: unwrap ProcessorMixin to raw tokenizer for text-only SFTTrainer on VLM-architecture models 2026-02-18 19:13:20 +00:00
Roland Tannous
a243ea411d Merge pull request #154 from unslothai/feature/colab-notebook 
Add Google Colab Support for Unsloth Studio
 2026-02-18 23:02:22 +04:00
Roland Tannous
a6e2fa5b3a Merge remote-tracking branch 'origin/nightly' into fix/dataset-mapping-vlm-text-datasets 2026-02-18 18:11:45 +04:00
Roland Tannous
940328ce1e Merge remote-tracking branch 'origin/nightly' into feature/colab-notebook 2026-02-18 17:41:07 +04:00
Roland Tannous
fb556d9a2c Merge branch 'fix/sm_120-flex-attention-temp-disable' into nightly 
renamed UNSLOTH_FLASH_ATTENTION to UNSLOTH_ENABLE_FLASH_ATTENTION to match actual environment variable in
unsloht
 2026-02-18 09:24:52 +00:00
Roland Tannous
d57b2742ab renamed UNSLOTH_FLEX_ATTENTION to UNSLOTH_ENABLE_FLEX_ATTENTION 2026-02-18 09:21:53 +00:00
Roland Tannous
ee5236228d Merge pull request #152 from unslothai/fix/sm_120-flex-attention-temp-disable 
Disable flex attention on Blackwell+ GPUs (sm_120+) at startup
 2026-02-18 13:00:43 +04:00
Roland Tannous
5a02ed4f0f Disable flex attention on Blackwell+ GPUs (sm_120+) at startup 2026-02-18 08:58:25 +00:00
Roland Tannous
346d7cd13a Merge pull request #151 from unslothai/fix/trainer-hang-resource-cleanup 
Fix/trainer hang resource cleanup
 2026-02-18 12:39:40 +04:00
Roland Tannous
d69431fa57 Scale dataset num_proc dynamically to cpu_count//3 instead of hardcap 8 2026-02-18 08:38:53 +00:00
Manan17
76cd1dc24c fixing the hangup of training after multiple back to back training processes 2026-02-18 08:18:13 +00:00