forum

1495 threads · Page 9 of 30

Offline resources 12 messages
Cannot run trainer.train() — backward-graph RuntimeError 4 messages
[Support Request] Need Working Code for Original Qwen/Qwen2.5-VL-7B-Instruct Model Fine-tuning 14 messages
Not able to run unsloth gemma 3 bnb on with dynamic lora on vllm? 10 messages
Open
GRPOTrainer compute_loss function shape mismatch for matmul 2 messages
multi gpu 5 messages
GRPO Standby H100 7 messages
help with test train split 17 messages
Gemma 3n Training Notebook Fail 6 messages
Solved
ValueError: Supplied state dict for model.layers.0.mlp.experts.down_projs.0.weight does not contain 3 messages
finetuning SmolLM-135 for coding generation 2 messages
Open
Unsloth with 9070 XT in Fedora 42 (ROCm 6.3.1.1) 4 messages
Support for Gemma 3N model vLLM support 3 messages
FINETUNE unsloth/gemma-3-4b-it but result really badddddd 2 messages
Open
Unsupported: Unsupported Tensor.requires_grad_() call 2 messages
Cannot finetune Qwen3Next 4bit BnB -- Compatibility issues 14 messages
Open
regarding your new 7 messages
How to merge multiple LoRA adapters (vision + text) for combined inference in Unsloth? 71 messages
load_in_8bit and KeyError: "" 3 messages
Finetuning Gemma 3 4B/12B/27B on multi-image datasets 3 messages
cuda version 17 messages
How to calculate F1 Score and Accuracy when training with Unsloth? 7 messages
Open
VLLM OOMs on small models. 6 messages
Open
gpt oss 20b and KeyError: '' 28 messages
Does Dora work with Unsloth? 2 messages
Gemma-3 1B - Saving a GGUF results in a bad model 108 messages
code says to try increasing max_seq_length but that fixes nothing 4 messages
Fine-tuned unsloth/gemma-3-1b-pt model produces gibberish/empty output after quantization (GPTQ/AWQ/ 33 messages
unable to push to hf 3 messages
Difference between training with 'load_in_4bit' vs a pre quantized bnb-4bit model 16 messages
What sampling settings does unsloth use during training 2 messages
Open
error when finetune Qwen2.5-vl-7b, but it's fine in gemma-3-4b-it and gemma-3-12b-it 3 messages
RuntimeError: Boolean value of Tensor with more than one value is ambiguous 6 messages
Solved
AttributeError: module 'transformers.models.bit.modeling_bit' has no attribute 'Linear' 17 messages
'Qwen2Config' object has no attribute 'text_config' 5 messages
unsloth trainer code work 13 days ago, now same code give runtime error: boolean value of tensor... 10 messages
Bots only? 3 messages
Sequence classification notebook no longer works: 20 messages
Does Unsloth have quantized versions of optimizers other than AdamW? 10 messages
Open
there is an error when finetuning gemma-3-12b-it by lora 13 messages
Just chat with bot 41 messages
Gemma support 15 messages
Help needed: Converting fine-tuned llama3.2-vision model to GGUF format 3 messages
gpt-oss: can't load local checkpoint 6 messages
Saving GPT-OSS as .GGUF 13 messages
Help Needed: Running Llama3.2 (11B) Vision Notebook on A100 Runtime 5 messages
Issue with LoRA Merge Inference on L4 Machine Using Unsoth 8 messages
Temporary - just to use the bot 8 messages
Open
[Bug] Garbled text output likes ה during inference when UNSLOTH_VLLM_STANDBY=1 is enabled using Qwen 3 messages
Is it possible to update a GRPO notebook to use LLM feedback? 2 messages