Daniel Han comments

Results 781 comments of


                                            Daniel Han

Can we suport qwen series?

Oh I think someone made a PR for Qwen

Errors on fine tuning gemma

I'm assuming you forgot to call `FastLanguageModel.get_peft_model`

Suggestion for improved clarity of supported torch versions

Oh we ask people now to install `!pip install --no-deps "xformers

unsloth with python ver.3.12.3

Oh I don't think we support 3.12 yet since Xformers does not work on 3.12

Can't Train Phi3 on Kaggle (ValueError: Pointer argument (at 2) cannot be accessed from Triton (cpu tensor?))

Did you turn on the GPU?

Requesting support for IBM's OpenSource Granite models

Oh interesting!

Error in triton while running unsloth/mistral instruct v0.2

Oh thats a weird error - i will try Conda installs and get back to you

TypeError: MistralForCausalLM.forward() got an unexpected keyword argument 'causal_mask'

Hmm thats weird - do you know if Colab Mistral works?

Jinja2 Fails at 50%

Oh nice catch!! Did not know about Jinja2 - thanks for that - will add it in!

Lora merging on cpu?

Sorry on the delay! Oh it should dispatch to CPU on the fly! Try doing `model.save_pretrained_merged("model", tokenizer, save_method = "merged_16bit", maximum_memory_usage = 0.7)`