unsloth
Python Mediumunslothai/unsloth
60,135 stars
5,153 forks
100 open issues
Active Apr 2026
Beginner-Friendly Issues 100
Issues tagged for new contributors
bug
[Bug] save_pretrained_merged silently fails to dequantize FP8 Devstral model
#4919 · Apr 8, 2026
[Bug] Qwen3.5 MoE training issues
#4908 · Apr 8, 2026
Unrelated
Discussion
[Bug] Windows: User Names with spaces cause uv/pip to fail on install
#4904 · Apr 7, 2026
bug
[Feature] Unsloth studio: Set dataset streaming mode
#4903 · Apr 7, 2026
feature request help wanted good first issue
环境太严格了
#4898 · Apr 7, 2026
[Feature] add custom tag or name to model training
#4895 · Apr 7, 2026
feature request
bug
[Bug] Incorrect args for gradient_checkpoint_enable
#4886 · Apr 7, 2026
bug
[Bug] GRPO: grpo_compute_loss_slow called with wrong positional args
#4885 · Apr 7, 2026
bug
[Bug] Please fill in your issue title here.为什么我下载的beta版本不能训练哇？是本身就不支持吗？
#4871 · Apr 6, 2026
on roadmap feature request
[Docs] Korean documentation translation contribution
#4861 · Apr 5, 2026
[Bug] Can't train Qwen3.5 or Gemma4 on multimodal datasets in Unsloth Studio
#4859 · Apr 5, 2026
bug
[Feature] Multi-host support for Unsloth Studio
#4858 · Apr 5, 2026
feature request
feature request help wanted good first issue
feature request
[Feature ]Support for PyTorch 2.11 to catch up with the newest vllm
#4851 · Apr 4, 2026
feature request
fixed bug
Torch cpu error
#4848 · Apr 4, 2026
unsure bug? insufficient information
feature request
bug
[Bug] GGUF export failed: Unsloth [Unsloth Studio]
#4845 · Apr 4, 2026
bug
[Bug/Issue] Unable to make it work with Pascal architecture cards.
#4843 · Apr 4, 2026
feature request bug
feature request colab
[Feature Request] Support DevStral Small 2 on Transformers v5
#4832 · Apr 3, 2026
feature request
[Bug] Failed to load model: Non-relative patterns are unsupported
#4831 · Apr 3, 2026
bug
[Installation Issue]To optimize the installation process of PyTorch
#4830 · Apr 3, 2026
feature request
fixed
[Bug] llama.cpp build failed at: cmake configure
#4826 · Apr 3, 2026
fixed
bug
bug
[Bug] Unsloth Studio Installation Failure: macOS M4 (arm64)
#4774 · Apr 2, 2026
fixed
Discussion
[Bug] Shell injection in GGML export conversion
#4766 · Apr 2, 2026
bug
[Feature]Request: 4-bit Quantized Export (AWQ/GPTQ) for vLLM in Unsloth Studio
#4761 · Apr 1, 2026
feature request help wanted good first issue
[Feature] Draft model / speculative decoding
#4753 · Apr 1, 2026
feature request help wanted good first issue
currently fixing bug
[Bug] Different behaviour observed by using `TrainingArguments` vs `SFTConfig`
#4735 · Mar 31, 2026
Discussion
[Bug] Compatibility Issue with PyTorch 2.6.0 (Python 3.12) - recompile_limit removed
#4728 · Mar 31, 2026
Discussion
[Bug] Problem loading models offline in Unsloth.
#4727 · Mar 31, 2026
bug
[Bug] Win10 - unsloth CLI was not installed correctly.
#4702 · Mar 30, 2026
insufficient information
[Bug] Unsloth Studio installation fails on Windows 11 with scoop
#4701 · Mar 30, 2026
insufficient information
on roadmap feature request
[Bug] Qwen3.5 fine-tuning not working in Unsloth Studio
#4688 · Mar 29, 2026
[Feature] Support FireRed-Image-Edit
#4686 · Mar 29, 2026
feature request
[Bug] Out of Memory during multi-GPU training
#4685 · Mar 29, 2026
inactive
feature request
[Bug] lama-server binary in Docker image crashes with "Illegal instruction"
#4683 · Mar 29, 2026
unsure bug?
Multi-GPU DDP support for FastSentenceTransformer?
#4680 · Mar 28, 2026
feature request
feature request
[Feature] A way to view and change llama.cpp server args
#4672 · Mar 28, 2026
feature request
[Bug] Saving fine-tuned Qwen3.5 model yields text-only tokenizer
#4671 · Mar 28, 2026
bug
bug
First experience of using Unsloth Studio - Feedback
#4668 · Mar 27, 2026
bug
[Feature] Unsloth studio Export chat configs to ollama
#4660 · Mar 27, 2026
feature request
[Bug] GPU not being utilized properly for AMD in Unsloth Studio.
#4644 · Mar 27, 2026
bug
[Feature] Preliminary MultiGPU support not working properly in Unsloth Studio
#4643 · Mar 27, 2026
feature request
[Bug] Cannot train with images on a multi-model Qwen model
#4641 · Mar 27, 2026
bug
sm_120 arch forces SDPA attention, even with functional xformers
#4631 · Mar 26, 2026
bug
Error loading checkpoint: '-'
#4628 · Mar 26, 2026
Unrelated
[Bug] Failed to load model
#4626 · Mar 26, 2026
Unrelated
[Feature] Allow for using remote GPU with a client server model.
#4613 · Mar 26, 2026
feature request
[Feature] Add Continued Pretraining in Unsloth Studio
#4565 · Mar 25, 2026
feature request
[Bug] .save_pretrained_gguf failing issue on databricks
#4557 · Mar 24, 2026
feature request unsure bug?
Request: Add Support for OLMo3 (OLMo2 Already Supported)
#4546 · Mar 24, 2026
feature request help wanted good first issue
[feature request] [studio] [windows] introduce support for visual studio tools 2026
#4544 · Mar 24, 2026
feature request
[Bug] unsloth / vllm / gpt-oss-20b
#4541 · Mar 23, 2026
[Feature] Unsloth/ Whisper/Large-v3 - S3 Bucket connection
#4539 · Mar 23, 2026
feature request help wanted good first issue
bug
[Bug] GGUF export broken for Qwen3.5-9B merged checkpoint (text-only fine-tune)
#4534 · Mar 23, 2026
bug
bug
feature request
[Bug] Win11 - unsloth CLI was not installed correctly.
#4520 · Mar 22, 2026
bug
[Bug] High CPU usage during training。
#4519 · Mar 22, 2026
bug
[Bug] Unsloth Studio Colab doesn't give the correct URL
#4516 · Mar 21, 2026
bug
Feature Request: Skip CMake installation flag and allow manually adding LlamaCPP
#4514 · Mar 21, 2026
[Bug] W11 25H2 "Failed to install" (maybe due to multiple RTX in the system ?)
#4513 · Mar 21, 2026
unsure bug?
[Bug] Studio didn't use local model file though I have a local cache.
#4508 · Mar 21, 2026
bug
[Feature] configurable number of threads Llama.cpp parameter
#4506 · Mar 21, 2026
feature request
bug
bug
[Feature] integration of Q-Galore
#4499 · Mar 21, 2026
feature request
Geplante ROCm-Unterstützung für Radeon 8060S?
#4498 · Mar 21, 2026
[Feature] Add llama.cpp vulkan backend support for non-nvidia cards
#4497 · Mar 21, 2026
feature request
[Feature] ROCm Support in Unsloth Studio
#4496 · Mar 21, 2026
feature request
[Feature] Custom + system variables @ system prompt
#4495 · Mar 21, 2026
feature request
[Bug] missing unsloth executable after install
#4488 · Mar 20, 2026
bug
[Bug] Windows 11 Installation Issues.
#4487 · Mar 20, 2026
bug
[Bug] completion_only_loss=True.
#4486 · Mar 20, 2026
bug
DDP Support for Vision-Language Model Fine-Tuning
#4485 · Mar 20, 2026
bug
[Feature] Add Qwen3 VL embedding SFT notebook
#4481 · Mar 20, 2026
feature request
Simplified Chinese documentation has inaccurate translations and unclear wording
#4480 · Mar 20, 2026