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
FSDP2: gather for push_to_hub, refuse GGUF / torchao exports, size offloaded state and 4-bit counts per shard (#14205)
fbe440f Daniel Han committed 1 month ago
gpt-oss MXFP4 without triton_kernels on transformers 4.x: keep dense expert biases in the model dtype (#14275)
c73389c Daniel Han committed 1 month ago
GatedDeltaNet fla op engages in training; large-head self-check survives a non-reentrant checkpoint (#14233)
745bdb2 Daniel Han committed 1 month ago
QLoRA quant-state fallback: gate per weight, not a process-global flag (#14231)
2ff041b Daniel Han committed 1 month ago
Studio: accept a quoted local training base and name single-file checkpoints in the refusal (#14236)
9836ddc Daniel Han committed 1 month ago
Bump install.sh / install.ps1 pins to unsloth>=2026.10.7, unsloth-zoo>=2026.10.7 (#14276)
0a96b9d Daniel Han committed 1 month ago
Tests: start every diffusion_segment test with no GPU arbiter resident (#14271)
e680e30 Daniel Han committed 1 month ago
Studio: a chat remembers the model it last used, not only the one it started on (#14259)
87c6408 Daniel Han committed 1 month ago
unsloth start vscode: print the group instead of writing a dead file inside a plain container (#14237)
d6a7ae8 Daniel Han committed 1 month ago
Format the two files #14247 left off the formatter fixed point (#14270)
f210636 Daniel Han committed 1 month ago
Studio: timestamp every other sampled video frame for GGUF chat (#14254)
f7ec00f Daniel Han committed 1 month ago
Tests: make image-engine-choice and image-mask-from-text typecheck (#14272)
bf22118 Daniel Han committed 1 month ago
Tests: echo gguf_filename in the image text encoder Playwright status stub (#14269)
6c15981 Daniel Han committed 1 month ago
Tests: a grandchild reaped while its /proc stat is read counts as gone (#14268)
21c0205 Daniel Han committed 1 month ago
Unsloth Studio / Desktop: name chat exports after model, title and day, and reopen the last save folder (#12586)
7790e1b Leo Borcherding committed 1 month ago
Studio: eject a model with a running chat from the Loaded models card (#14161)
4ea4c75 Nilay committed 1 month ago
Studio: let an approved rm run, never a recursive one (#14238)
28a222f Daniel Han committed 1 month ago
MFU logging under fullgraph, NCCL logs under distributed debug, token-budget metrics across ranks (#14207)
9f78f5e Daniel Han committed 1 month ago
Decode: cover explicit positions past the lazy RoPE cache in Gemma2, Cohere, Granite and FalconH1 (#14199)
7e3e0d9 Daniel Han committed 1 month ago
Studio: a kept load short on its own GPU no longer unloads models on other GPUs or on the CPU (#14234)
c708b1d Daniel Han committed 1 month ago
Context parallel: wrap trainer methods only once a trainer enables CP (torch 2.10 step time) (#14152)
40e2001 Daniel Han committed 1 month ago
Studio: show the end of long prompts in the API monitor (#14241)
2c7c841 Daniel Han committed 1 month ago
Studio installer: show the previous unsloth version next to the new one (#14248)
5278547 Daniel Han committed 1 month ago
Studio: make inpaint and localized-edit masks from text with SAM 3 (#14235)
bff0a25 Daniel Han committed 1 month ago
Studio: raise the startup bundle budget for #11937 (#14264)
80e2140 Daniel Han committed 1 month ago
Studio: wire in DeepSeek-V4.1-Flash-1017 (deepseek41) (#14260)
d1343c4 Daniel Han committed 1 month ago
Studio Images: per-load engine choice for GGUF picks (native sd.cpp for LoRA) (#14232)
6ccf0e4 Daniel Han committed 1 month ago
Studio: delete selected recipe blocks with the Delete key (#14223)
f5e91f1 Daniel Han committed 1 month ago
CI: keep one generation of prefix-restored caches, share Core's pip cache (#14263)
e632df1 Daniel Han committed 1 month ago