Search results

[Feature] Please add an option to Unsloth Export that disables slicing or defines shard sizes

feature requestFeature request pending on roadmapFeature request pending on roadmap

Status: Open.

#6083 In unslothai/unsloth;

· victorxstc opened on Jun 7, 2026

[Bug] GGUF export unnecessarily copies finished GGUF files to .unsloth directory in C: drive, regardless of the local source/destination addresses set by the user. This also causes WinError 112 on systems low on free space on drive C:

bug

Status: Open.

#6082 In unslothai/unsloth;

· victorxstc opened on Jun 7, 2026

[Feature] I need voice input and output in my language

feature requestFeature request pending on roadmapFeature request pending on roadmap

Status: Open.

#6081 In unslothai/unsloth;

· Damian-o2 opened on Jun 7, 2026

[Bug] Unsloth Studio should NOT SILENTLY and automatically download the Helper LLM. It should ask the user on its GUI, giving the user options for various scenarios.

bug

Status: Open.

#6080 In unslothai/unsloth;

· victorxstc opened on Jun 7, 2026

[Feature] Please add 2 different "Browse" buttons for selecting local Source and Destination folders, in the Export part of Unsloth Studio

feature requestFeature request pending on roadmapFeature request pending on roadmap

Status: Open.

#6078 In unslothai/unsloth;

· victorxstc opened on Jun 7, 2026

Does plain, lossless conversion from safetensors to GGUF (without any quantization), using Unsloth Studio, make the inference any faster?

Status: Open.

#6077 In unslothai/unsloth;

· victorxstc opened on Jun 7, 2026

Custom Providers

feature requestFeature request pending on roadmapFeature request pending on roadmap

Status: Open.

#6074 In unslothai/unsloth;

· yassenehrenmann1 opened on Jun 7, 2026

[Feature] runtime context length on hub model

feature requestFeature request pending on roadmapFeature request pending on roadmap

Status: Open.

#6073 In unslothai/unsloth;

· james5635 opened on Jun 7, 2026

[Bug] Qwen3.5 / Qwen3.6 GGUF export pipeline truncates trailing 'ssm_conv1d.weight' tensors in final layers

bug

Status: Open.

#6071 In unslothai/unsloth;

· NighmareGit opened on Jun 7, 2026

[BUG] Rebuilt (post-June-5) Gemma-4 GGUFs still carry broken text-tower weights — verified at the tensor level, engine-independent. PPL 192.9 vs true 4.68

bug

Status: Open.

#6056 In unslothai/unsloth;

· nihilistau opened on Jun 7, 2026

[Bug] Too much download for a MINOR Unsloth Studio update -- llama.cpp installation fell back to source build

bug

Status: Open.

#6053 In unslothai/unsloth;

· victorxstc opened on Jun 6, 2026

[Bug] Role "System" in API calls

bug

Status: Open.

#6050 In unslothai/unsloth;

· SergejMuhic opened on Jun 6, 2026