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
Repair the four a-k checks the Qwen-Image-2.1 wave left red (#11573)
023730c Daniel Han committed 2 months ago
Hold the escape click guard until the button comes up (#11568)
be072ee Michael Han committed 2 months ago
Keep weight_decay when embedding_learning_rate is set (#11470)
ed004ac Vineeth Sai Varikuntla committed 2 months ago
Bump install.sh / install.ps1 pin to unsloth>=2026.9.9 (#11570)
bbfabb2 Daniel Han committed 2 months ago
Studio: finish an update with the installer it installs (#11566)
6f6d49a oobabooga committed 2 months ago
Studio: install the pinned Diffusers commit from a zip when there is no git (#11565)
f394701 Daniel Han committed 2 months ago
Studio: drag sidebar rows in the desktop app, and drop the drag hint (#11561)
92e1485 Michael Han committed 2 months ago
Stub the low-disk check in the tool-approval harness (#11552)
ffbc3a5 Daniel Han committed 2 months ago
Bump install.sh / install.ps1 pin to unsloth>=2026.9.8 (#11550)
541dd1a Daniel Han committed 2 months ago
Unsloth Studio: give the memory planner the companion split on a pipeline image load (#11548)
f2e6a44 Leo Borcherding committed 2 months ago
Studio: do not engage FBCache on a transformer that reuses a prefix KV cache (#11546)
c23f8b0 Daniel Han committed 2 months ago
feat(chat): make agentic turns survive tab-close (#10365)
0270575 CelesteHeartsong committed 2 months ago
Studio: reuse one database connection per thread for durable chat generation runs (#11525)
d4a402b Daniel Han committed 2 months ago
Fix push_to_hub_merged for full fine-tunes on transformers 5 (#11482)
76b06af Nilay committed 2 months ago
Merge branch 'main' of https://github.com/unslothai/unsloth
868039c Daniel Han committed 2 months ago
Pin the Dynamic 2.0 4-bit text encoder for Qwen-Image-2.1's native route (#11542)
71912e8 Daniel Han committed 2 months ago
Studio: take the hosted pre-cast fp8 text encoder by default on Qwen-Image-2.1 (#11539)
0fb5edc Daniel Han committed 2 months ago
Load the published Qwen-Image-2.1 INT8 checkpoint, and four more follow-ups (#11535)
67bc857 Daniel Han committed 2 months ago
Make the thinking trace read as a trace, not as the answer (#11541)
75dba26 Michael Han committed 2 months ago
Make the GPU probe harness name the helper it failed to extract (#11538)
5b6b7c3 Daniel Han committed 2 months ago
Read the packing blocker's branch by behaviour, not by its spelling (#11534)
c9e461a Daniel Han committed 2 months ago
Make main formatter-clean so pre-commit stops rewriting every PR (#11540)
34094a4 Daniel Han committed 2 months ago
Load prequantized multimodal checkpoints correctly on transformers 5.4 and 5.5 (#11452)
151517f Daniel Han committed 2 months ago