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 tests: forget the managed provider URL setting cache around each test (#12288)
7b2cb03 Daniel Han committed 1 month ago
Studio: keep working when GitHub rate limits or is down (#10670)
b84349a Nilay committed 1 month ago
Train LoRA on NVFP4 compressed-tensors checkpoints with the weights kept packed (W4A16) (#12185)
3cffce2 Daniel Han committed 1 month ago
Keep what you typed with a document when a long chat is compacted (#12238)
99318b0 Nilay committed 1 month ago
Studio: stop the stall watchdog from killing Xet downloads that are still receiving data (#11594)
249e537 Nilay committed 1 month ago
Qwen3.5: skip compiled regions on eager decode steps, make CUDA graph decode opt-in (#12213)
34f3455 Daniel Han committed 1 month ago
feat(studio): run fp16 Laya checkpoints in fp16 on MLX (#12256)
90db6a8 Long Yixing committed 1 month ago
Raise the minimum typer version so the unsloth command starts (#11148)
099e8e8 Nilay committed 1 month ago
fix(prompt storage): follow up, restore prompt list bookmarking, the kind pill and drag autoscroll (#9788)
591601b Leo Borcherding committed 1 month ago
Unsloth Studio / Desktop: add Canvas's console errors, and offer them to the model (#11744)
fea95db Leo Borcherding committed 1 month ago
Fix DDP "marked ready twice" for VLMs with CPU offload + TiledMLP (#4240)
46764a6 Daniel Han committed 1 month ago
Studio: keep pop-up messages off the Run settings panel (#11486)
8b8306b Nilay committed 1 month ago
Unsloth Studio (AMD RDNA 1, Windows): install ROCm torch for the RX 5700 XT from AMD's multi-arch index (#11755)
79ba3ea Leo Borcherding committed 1 month ago
Studio: save the chat template a base model was trained with (#12240)
419f447 Nilay committed 1 month ago
Decline the fused LoRA kernels under FSDP, where the weights are shard views (#11132)
eaa56d6 Daniel Han committed 1 month ago
Studio: read attached HTML files in the encoding they declare (#12235)
052b6e9 Nilay committed 1 month ago
Studio: reopen the artifact panel after it has been dragged shut (#12278)
0cc9ed2 Daniel Han committed 1 month ago
Unsloth Studio / Desktop (AMD): explain why an NVIDIA+AMD host installed with `CUDA_VISIBLE_DEVICES=""` shows no GPU (#12249)
fe29f23 Leo Borcherding committed 1 month ago
Unsloth Studio / Desktop (AMD): "Automatic" llama.cpp follows the installed torch on a mixed NVIDIA+AMD host (#12250)
d3ca2b0 Leo Borcherding committed 1 month ago
Studio: hide the stray scrollbar under the composer at fractional zoom (#12279)
43c0a41 Michael Han committed 1 month ago
fix: handle zero-strided tensors in fast_rope_embedding (#3781) (#4233)
42cffd5 Daniel Han committed 1 month ago
refactor(trainer): add warning for ignored eval_steps (#4225)
1c67252 Daniel Han committed 1 month ago
Unsloth Studio / Desktop: allow text-encoder nvfp4 on pre-Blackwell GPUs and explain TE precision refusals (#11413)
443dfda Maheswar Kumar committed 1 month ago
Studio: accept the chat columns the trainer accepts in the dataset check (#11849)
4c42e20 Nilay committed 1 month ago
ci: enforce npm ci on Studio install paths (follow-up to #5604) (#5616)
fbccb93 Daniel Han committed 1 month ago