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 diffusion: pin Triton libdevice before the first compile so restarts hit the FX graph cache (#13713)
8267225 Daniel Han committed 1 month ago
Unsloth Studio: a JSONL reply that quotes <tool_call> no longer stalls the tool parser and fails the turn with RecursionError (#14097)
cfe31dc Leo Borcherding committed 1 month ago
fix: preserve padding-free DPO sample boundaries in fast Llama/Qwen2 (#14251)
7f4736d cherry77-cloud committed 1 month ago
Studio: API keys run tools in the sandbox; Full access needs the Unsloth UI (#14345)
17ecb2e Daniel Han committed 1 month ago
Studio: honor width, height, steps, guidance, seed and negative_prompt on /v1/images/generations (#14351)
e9127c9 Daniel Han committed 1 month ago
Studio: open each Settings tab at the top instead of the previous tab's scroll offset (#14353)
9c5b09b Daniel Han committed 1 month ago
unsloth start (Windows): pass the sub-agent task over stdin so non-npm .cmd shims don't mangle it (#14300)
705bcc0 Leo Borcherding committed 1 month ago
fix(studio): keep download approval and saved filenames aligned (#14278)
ca02529 Nilay committed 1 month ago
Studio: skip causal-conv1d for Qwen3.5 / 3.6 / Qwen3-Next when unsloth-zoo covers decode (#14148)
670d9b8 Daniel Han committed 1 month ago
fix(studio): align Whisper training labels with decoder prefix (#14244)
ce76346 cherry77-cloud committed 1 month ago
Normalize role names when converting ChatML to Alpaca (#14304)
e6bfccd ShryukGrandhi committed 1 month ago
Studio: preserve browser panel drafts across tab switches (#14280)
3c8ffe3 Nilay committed 1 month ago
Studio: keep runtime flash-attn installs wheel-only (#14287)
9c2ee22 oobabooga committed 1 month ago
fix(studio): MLX memory estimate no longer allocates the model, and sizes DeepSeek V4 / V4.1 (#14325)
0753958 Long Yixing committed 1 month ago
Studio diffusion: multi-GPU component placement and CFG-parallel (#13725)
811c08e Daniel Han committed 1 month ago
OpenAPI loop test: build the real schema outside the measured window (#14346)
7ac1d2b Daniel Han committed 1 month ago
PEFT MoE conversion snapshot: add the Qwen3-Omni MoE text type and thinker aliases (#14344)
c200c08 Daniel Han committed 1 month ago
Prebuilt wheels: build causal-conv1d and mamba-ssm for Linux torch 2.10 to 2.12 (#14336)
4b8447f Daniel Han committed 1 month ago
Move the Unsloth Studio MLX pins to mlx-vlm 0.7.9 (#14307)
3511e95 Long Yixing committed 1 month ago
Desktop: offer a chat-only (GGUF) install and keep it through repairs (#14245)
3ceb606 Daniel Han committed 1 month ago
Studio: keep project sources when a knowledge base is picked (#14255)
e3b00a1 Daniel Han committed 1 month ago
Distributed tests: pick MASTER_PORT outside the ephemeral range, write vision results atomically (#14334)
846cc5e Daniel Han committed 1 month ago
Point Unsloth docs links at unsloth.ai/docs instead of docs.unsloth.ai (#14333)
d9aded3 Michael Han committed 1 month ago
Raise the startup bundle budget for the modern toggles (#14322) (#14331)
617844a Daniel Han committed 1 month ago
Studio: modern toggles, 34 x 20 with a flat centred thumb (#14322)
37f5374 Michael Han committed 1 month ago