unsloth · Issues· 1222 open
Open on GitHubLocally synced open issues (discussions stay on GitHub)
- #11143
[Bug] b10995-mix-3e83366: Qwen3.8-Flash-Next MTP aborts at load (nextn.hc_head_norm still [hc_dim] after rebase)
Updated Sep 17, 2026 - #4858
[Feature] Multi-host support for Unsloth Studio
feature requestUpdated Sep 17, 2026 - #9117
[Feature]We hope to add new model download sources
feature requestUpdated Sep 17, 2026 - #11189
[Feature] Load model button in API board
feature requestUpdated Sep 17, 2026 - #11187
[Bug] Studio not starting on 0.0.0.0
feature requestbugUpdated Sep 17, 2026 - #11186
[Bug] "Thread __LOCALID_s9k2Cx0 was not persisted" when uploaded documents.
feature requestbugUpdated Sep 17, 2026 - #2613
[Bug] Full Finetune: Tensors of floating point dtype can require gradients
unsure bug?Updated Sep 17, 2026 - #11184
[Feature] Implement Vulkan Training/Fine-tuning
feature requestUpdated Sep 17, 2026 - #11140
OpenAI API: auto-switch does not cold-load a downloaded GGUF, so `/v1/chat/completions` returns 400 "No model loaded" when nothing is loaded
Updated Sep 17, 2026 - #6730
Feature Request: Support MiCA (Minor Component Adaptation)
feature requestUpdated Sep 17, 2026 - #10397
Make SSH restrictions consistent and allow approved deployments
Updated Sep 17, 2026 - #10657
[Bug] Unsloth installer does not recognize ROCm 7.14 and installs PyTorch 2.11 with ROCm 7.2.
feature requestbugUpdated Sep 16, 2026 - #8529
[Bug] RX 5700XT doesn't recognize in Unsloth Desktop
bugAMDUpdated Sep 16, 2026 - #11141
Studio: expose prefill progress over the API so clients can show the wait before the first token
Updated Sep 16, 2026 - #11135
Unsloth Desktop packaged for Nixpkgs
Updated Sep 16, 2026