Commit graph

74 commits

Author SHA1 Message Date
Shine1i
f6397bf1ac feat: add cancelation support for chat generation and streaming tasks 2026-02-15 18:23:27 +01:00
Shine1i
78c7b6d7ba fix lora: outputs path local 2026-02-15 16:58:24 +01:00
Roland Tannous
79ee9f6f2d Merge pull request #96 from unslothai/feature/inference-yaml-ordered
Added the inference defaults for models
2026-02-15 17:43:31 +04:00
sshah229
2483b98985 added the inference fetching from model mappers 2026-02-15 02:48:53 -07:00
Manan17
e672458821 Adding save-steps to the SFTConfig 2026-02-15 09:37:54 +00:00
sshah229
e6bd0a20bb added default inference config for default.yaml 2026-02-15 02:13:24 -07:00
sshah229
a105e60d30 added default inference config from unsloth notebooks 2026-02-15 02:13:24 -07:00
sshah229
81be5dc080 added configs from Ollama 2026-02-15 02:13:24 -07:00
sshah229
9b5e02029e added inference defaults from unsloth guides 2026-02-15 02:13:24 -07:00
Roland Tannous
c866656502 Merge pull request #29 from unslothai/feature/export
Added the export routes and pydantic models
2026-02-15 12:42:01 +04:00
Roland Tannous
6ff2403356 Merge pull request #83 from unslothai/fix/training-stuck-multiprocessing-cuda
Fixing stuck training processes
2026-02-15 10:18:49 +04:00
Roland Tannous
9195311b1c fix: add no-cache headers to index.html to prevent stale frontend after rebuild 2026-02-15 06:06:27 +00:00
Manan17
df5f45058f Fixing stuck training processes 2026-02-15 05:38:06 +00:00
Manan17
354b7d0aca feat: add cancel or save and stop training 2026-02-15 00:00:22 +00:00
Roland Tannous
ac8128519d decouple reliance of backend on frontend for is_lora 2026-02-14 20:13:50 +00:00
Roland Tannous
4399687f93 strip extra debug statements 2026-02-14 19:23:51 +00:00
Roland Tannous
4d868e8d2b replace model unloading and peft loading mechanism for compare feature 2026-02-14 19:18:49 +00:00
Roland Tannous
225b3f1750 del model.peft_config instead of using model.delete_adapter 2026-02-14 17:32:15 +00:00
Roland Tannous
f122154cf3 added print statements for activate_lora_adapter 2026-02-14 17:25:37 +00:00
Roland Tannous
754ccf1a67 swipped logger for print statements as logger isn't propagating 2026-02-14 17:21:26 +00:00
Roland Tannous
b930a17b1d added logging 2026-02-14 17:09:07 +00:00
Roland Tannous
9bee0a3f63 exclude default from model.delete_adapter 2026-02-14 17:03:52 +00:00
Roland Tannous
0b305fd822 _apply_adapter_state now calls revert_to_base_model and activate_lora_adapter properly 2026-02-14 16:57:24 +00:00
Roland Tannous
8403bac48d feat(inference): add use_adapter field for per-request adapter toggling in compare mode 2026-02-14 14:52:13 +00:00
Roland Tannous
4ab8f81780 migrate _generate_vision_response to use TextIteratorStreamer + background thread 2026-02-14 09:30:32 +00:00
Roland Tannous
480418b595 feat(inference): accept OpenAI multimodal content parts (image_url) in /chat/completions 2026-02-14 09:06:25 +00:00
Roland Tannous
8304060b9f fix: increase SSE progress timeout to 30min and allow step-0 updates 2026-02-14 05:47:22 +00:00
Roland Tannous
78e92346fc fix path in run_server 2026-02-14 05:06:17 +00:00
Roland Tannous
a505d3cf98 change default frontend path in run.py to studio/frontend/dist 2026-02-14 05:02:26 +00:00
Roland Tannous
286d5ff0a7 feat: wire custom_format_mapping through training pipeline to format_and_template_dataset 2026-02-13 21:07:36 +00:00
Roland Tannous
5652591d03 fix: auto-detect LoRA adapters for both local and remote HF models in ModelConfig 2026-02-13 20:54:40 +00:00
Roland Tannous
4e4fc367b6 fix: auto-detect multimodal datasets in /check-format without requiring is_vlm flag 2026-02-13 17:29:39 +00:00
Shine1i
0ebfb5be76 feat: add support for serialized previews in dataset API and improve training initialization logging 2026-02-13 13:47:17 +01:00
Roland Tannous
e9bf6ae368 read external ips with fallback to standard notation 0.0.0.0 2026-02-13 10:28:20 +00:00
Roland Tannous
afa7b3657f feat: show external IP in startup banner 2026-02-13 10:23:12 +00:00
Roland Tannous
45973a09e9 fix: rollback auth.db user row if token generation fails during setup 2026-02-13 10:11:07 +00:00
Roland Tannous
88f68eaadf refactor: remove gradio dependency from training backend 2026-02-13 09:25:49 +00:00
Roland Tannous
6beddf9f9e fix: change epoch type from int to float to match TrainerState 2026-02-13 06:51:55 +00:00
sshah229
ae0b809adf fixed the script directory 2026-02-12 21:55:36 -07:00
Roland Tannous
c78cb11f81 feat: add OpenAI-compatible POST /chat/completions endpoint with streaming and non-streaming support 2026-02-12 19:00:05 +00:00
Roland Tannous
d17c1b99d8 feat: add SSE reconnection resilience with spec-compliant event fields, Last-Event-ID resume, and metric_history fallback in /status 2026-02-12 17:58:48 +00:00
Roland Tannous
a2c7f1b1b8 return raw preview samples on format detection failure for manual column mapping 2026-02-12 15:39:56 +00:00
Roland Tannous
f780b0db43 feat(datasets): check-format to return preview samples 2026-02-12 11:25:47 +00:00
sshah229
e4b073985f added router in main 2026-02-11 18:51:53 -07:00
sshah229
46cc71f310 added the pydantic models and routes for export 2026-02-11 18:34:12 -07:00
Roland Tannous
a6ee9ee957 use get_device() for device selection and clear_gpu_cache() for GPU memory cleanup in inference, trainer, and export 2026-02-11 16:56:52 +00:00
Roland Tannous
8395652f59 added @needs_torch to test_cuda_oom 2026-02-11 16:12:37 +00:00
Roland Tannous
b618ea1f9f add @needs_mlx decorator on tests 2026-02-11 16:10:22 +00:00
Roland Tannous
e31d4a3c40 replace torch MPS with MLX 2026-02-11 16:04:35 +00:00
Roland Tannous
be02fac439 reset DEVICE type on fastapi lifespan exit 2026-02-11 15:58:13 +00:00