forum
In the gpt-oss notebook, the finetuned model doesn't seem to use French for reasoning?
Errors on latest release
Loss function to use for JSON outputs.
Open
gpt-oss-20b continued pretraining
Unsloth Index Out of Bound
Open
Error importing unsloth
Open
modules_to_save with vllm
Missing formatting_func error
Open
AcceleratorError: CUDA error: CUDA-capable device(s) is/are busy or unavailable
Solved
Creating Dynamic 2.0 GGUFs for custom models
Open
How to save a full fine tuned model
Issues with GLM-4.5-AIR
Open
Feature Request: Sequential Multi-Turn Finetuning Support
Open
Continued pretraining and training loss for Gemma 3
Open
any guide for train base to chat llm?
Medgemma 4b Vision
got ZeroDivisionError with qwen3 train_on_responses_only setup
Solved
Unable to DPO training and GGUF saving.
Open
Function is missing in dataloader "fetch_video"
gpt-oss-20B runtime error
Open
I want to provide computing resources
Bug when apply_chat_template with unsloth/Mistral-Small-3.2-24B-Instruct-2506
No module named 'transformers.models.gpt'
AttributeError: 'NoneType' object has no attribute 'shape'
Open
Unsloth doesn't detect rtx5060 on Windows
Solved
Loading unsloth model after fine-tuning.
Does Unsloth use HF Transformers, or llama.cpp for inference
Model is not trained when converted to GGUF/SFTrainer recommended settings for Gemma 3 1b (Q&A Data)
gpt-oss-120b ollama template
llama.cpp optimization help for gpt-oss-120b
Suggestion for Mistral Small 3.2
GPT-OSS sinks with Flex attn for LC
BackendCompilerFailed
Solved
Loading a Merged Gemma 3-270b-it model
Open
Gemma3 notebook error
SFT and GRPO on local RTX 3090
Are vision models supported for fast_inference=True?
Fine tuning
Open
Size mismatch after training unsloth_classification
vision model q8_0
How to properly retrain a model to a new language with tokenizer replacement?
Llama-3.2-11B
The size of tensor
no logs
Add proposed ReadMe for GLM 4.5 (Fixes MLX This Repo)
Solved
RuntimeError: Output 0 of UnslothFusedLossBackward is a view
Open
synthetic data notebook
performance low budget graphic card
How to load in a full fine tuned model?
Open
Version mismatch on latest libs and packs