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
Windows on ARM: take the wheelhouse from our own signed release (#11108)
c22f645 Daniel Han committed 2 months ago
Pin the win_arm64 scikit-learn to one version so the security audit cannot go red on its own (#11112)
907f198 Daniel Han committed 2 months ago
Keep an exported config's MTP declaration in agreement with the weights (#11042)
d7af91c Daniel Han committed 2 months ago
Docker: let UNSLOTH_STUDIO_PORT change Studio's port inside the container (#11045)
90cab2b Nilay committed 2 months ago
Studio: rewrite the guided tour and give every page one (#11105)
0fdb81a Michael Han committed 2 months ago
docker: let the quickstart generate the passwords and print them (#11109)
8e3a0e4 Daniel Han committed 2 months ago
Studio: stop merging the model twice when pushing a merged export to the Hub (#11050)
dba7490 Nilay committed 2 months ago
Studio: refine queue drag handles and resume icons (#11113)
5e80884 Michael Han committed 2 months ago
Studio: make the GGUF memory estimate match the compute buffers llama.cpp allocates (#11043)
820d377 oobabooga committed 2 months ago
Apply Q-GaLore weight decay before the projected update (#10931)
45124cc Vineeth Sai Varikuntla committed 2 months ago
Match stopping criteria separately for each generated sequence (#11039)
a1a38e9 Vineeth Sai Varikuntla committed 2 months ago
Studio: keep every turn when training a chat dataset with Alpaca format (#11053)
dc0a7db Nilay committed 2 months ago
docker: stop the toolkit installer reporting a healthy driver as missing (#11106)
d6dd50c Daniel Han committed 2 months ago
Windows on ARM: install the native ARM64 CUDA stack on NVIDIA hosts (#10282)
b535eca Daniel Han committed 2 months ago
studio: preserve transcripts and recall audio and image models (#10993)
f621d12 Maheswar Kumar committed 2 months ago
Studio: keep the GPU order the user asked for instead of re-emitting it ascending (#11034)
39268d3 Leo Borcherding committed 2 months ago
Studio: honor use_adapter on image turns in Base-vs-LoRA compare (#11047)
1f365b8 Nilay committed 2 months ago
Studio: stop hiding MCP tools that have a dot or a long name (#11051)
8453ef4 Nilay committed 2 months ago
Studio: stop context checkpoints from filling host RAM on hybrid recurrent models (#11073)
c92e450 oobabooga committed 2 months ago
Redact hf_token and wandb_token in `unsloth train --dry-run` output (#11100)
d0a5f4a Daniel Han committed 2 months ago
Keep an abandoned zoom call out of the next test's stub (#11090)
c0bdb04 Daniel Han committed 2 months ago
docker: correct four stale claims on the Hub page (#11099)
2aabc61 Daniel Han committed 2 months ago
Refine Studio auth card corners, shadow, and setup copy (#11095)
260e402 Michael Han committed 2 months ago
tests: route the uv requirements-path pwsh call through the shared runner (#11091)
1066d16 Daniel Han committed 2 months ago
Studio: send a download over Xet when HTTPS cannot fetch its largest file (#11061)
b28746f oobabooga committed 2 months ago
Studio: stop charging CPU-mapped input embeddings to the Metal context budget (#11060)
71c9a44 oobabooga committed 2 months ago
Studio: organize chat settings and simplify descriptions (#11088)
e00dde9 Michael Han committed 2 months ago
Studio: on low disk, load the largest GGUF quant that fits and show which one loaded (#11052)
ce66f60 Nilay committed 2 months ago