Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Python agentaichatgptdeepseekfine-tuninggemmaimage-generationllamallmllmsopenaipythonqwenreinforcement-learningself-hostedstable-diffusiontext-to-speechttsuiunsloth
Commits on main
Studio: add a tool output length setting to Settings > Chat (#13228)
67717d2 Daniel Han committed 1 month ago
Tests: fix three checks that fail for reasons other than a regression (#13265)
f0db334 Daniel Han committed 1 month ago
Studio (AMD): treat a multi-arch arch whose kernel pack is not installed as uncovered (#13231)
5d96b95 Daniel Han committed 1 month ago
Studio: keep the sidebar More flyout shut when a modal opens during its hover delay (#13237)
0ec5ae6 Daniel Han committed 1 month ago
README: point notebook-only Docker users at the core tag again (#13258)
3c152be Daniel Han committed 1 month ago
Chunked mask test: drop the now-unused _swap_function_references import (#13257)
151974c Daniel Han committed 1 month ago
Chunked mask test: unwrap every layer, including unsloth_zoo's mask wrapper (#13248)
c538b0f Daniel Han committed 1 month ago
Unsloth Studio / Desktop: export image LoRAs and pick them up again from a custom models folder (#13146)
6207fc8 Leo Borcherding committed 1 month ago
Studio: size a GPU cache for spilled MoE experts with --moe-cache-mib auto (#12951)
b15cda9 Daniel Han committed 1 month ago
Add opt-in tokenizers 1.0 release candidate encode path (#13214)
a5a992b Daniel Han committed 1 month ago
Studio: show context usage for a revisited chat under a cloud model or none (#13225)
d24f24d Daniel Han committed 1 month ago
Make Unsloth-patched tokenizers and processors picklable (#13209)
810efe4 Daniel Han committed 1 month ago
Let TRL AsyncGRPOTrainer train Unsloth models; fix GRPO vision rewards for tool images and TRL 1.15 (#13205)
4c902be Daniel Han committed 1 month ago
assistant_only_loss with Unsloth chat templates: fall back to response markers (#13204)
e7a1a91 Daniel Han committed 1 month ago
GRPO tool calling: accept Unsloth chat templates in TRL's response schema and training template lookup (#13202)
affbc0a Daniel Han committed 1 month ago
Support TRL 1.14 and 1.15: ORPO/CPO row cap, masked GRPO sampling metrics, cap to 1.15.0 (#13203)
3b865ba Daniel Han committed 1 month ago
Studio: pin Qwen-Image 2.1 Turbo directly under Qwen-Image 2.1 in the Images picker (#13239)
aed7dc4 Daniel Han committed 1 month ago
Studio: keep list numbers in attached Word and HTML files (#13185)
b939e54 Nilay committed 1 month ago
Studio: browser Downloads button matches the toolbar and closes on a second click (#13226)
04e9b26 Michael Han committed 1 month ago
Studio: keep heading text that sits inside a button on fetched pages (#13184)
aea22df Nilay committed 1 month ago
Unsloth Studio / Desktop (AMD, Windows): warn at startup when the AMD driver has the RDNA4 idle-eviction bug (#13208)
6884fb6 Leo Borcherding committed 1 month ago
Tests: drop duplicate tests and fix three that checked nothing new (#13211)
06c33f1 Leo Borcherding committed 1 month ago
Fall back to a dynamic KV cache when the static one would not fit (#13075)
8435b96 Daniel Han committed 1 month ago
Train text-diffusion LMs: DiffusionTrainer with recipes for DiffusionGemma, LLaDA, LLaDA-MoE, LLaDA2, Dream, Nemotron and SDAR (#13157)
cd14635 Daniel Han committed 1 month ago
Studio: keep tool calls and results when training on ShareGPT tool data (#13181)
0bdb7d5 Nilay committed 1 month ago
Studio: keep the chat composer typing after a send on WebKitGTK Wayland (#13222)
a89c7f5 Daniel Han committed 1 month ago
Studio (Jetson): keep JetPack's CUDA ahead of pip CUDA in llama-server and backend loader paths (#13191)
0512f3a Daniel Han committed 1 month ago