[BUG] Google colab fine-tune isn't working

[krecicki]

Prerequisites

• I have read the documentation.
• I have checked other issues for similar problems.

Backend

Colab

Interface Used

UI

CLI Command

No response

UI Screenshots & Parameters

I tried this link https://colab.research.google.com/github/huggingface/autotrain-advanced/blob/main/colabs/AutoTrain_LLM.ipynb#scrollTo=A2-_lkBS1WKA

I have used it before successfully.

NameError                                 Traceback (most recent call last)
[<ipython-input-4-bc00ccd290a9>](https://zjlogt075j9-496ff2e9c6d22116-0-colab.googleusercontent.com/outputframe.html?vrz=colab_20240405-060114_RC00_622131336#) in <cell line: 27>()
     25 weight_decay = 0.001 # @param {type:"number"}
     26 gradient_accumulation = 4 # @param {type:"number"}
---> 27 mixed_precision = bf16 # @param ["fp16", "bf16", "none"] {type:"raw"}
     28 peft = True # @param ["False", "True"] {type:"raw"}
     29 quantization = int8 # @param ["int4", "int8", "none"] {type:"raw"}

NameError: name 'bf16' is not defined

None of the options work. None of them are defined. Same error for all options.

I press the play button for the first to install the packages. I fill out all the information in the second block and hit play and get the error above.

I try running the third block and it just says this:

usage: autotrain <command> [<args>] llm [-h] [--text_column TEXT_COLUMN]
                                        [--rejected_text_column REJECTED_TEXT_COLUMN]
                                        [--prompt-text-column PROMPT_TEXT_COLUMN]
                                        [--model-ref MODEL_REF] [--warmup_ratio WARMUP_RATIO]
                                        [--optimizer OPTIMIZER] [--scheduler SCHEDULER]
                                        [--weight_decay WEIGHT_DECAY]
                                        [--max_grad_norm MAX_GRAD_NORM] [--add_eos_token]
                                        [--block_size BLOCK_SIZE] [--peft] [--lora_r LORA_R]
                                        [--lora_alpha LORA_ALPHA] [--lora_dropout LORA_DROPOUT]
                                        [--logging_steps LOGGING_STEPS]
                                        [--evaluation_strategy EVALUATION_STRATEGY]
                                        [--save_total_limit SAVE_TOTAL_LIMIT]
                                        [--save_strategy SAVE_STRATEGY] [--auto_find_batch_size]
                                        [--mixed-precision MIXED_PRECISION]
                                        [--quantization QUANTIZATION]
                                        [--model_max_length MODEL_MAX_LENGTH] [--trainer TRAINER]
                                        [--target_modules TARGET_MODULES] [--merge_adapter]
                                        [--use_flash_attention_2] [--dpo-beta DPO_BETA]
                                        [--chat_template CHAT_TEMPLATE] [--padding PADDING]
                                        [--train] [--deploy] [--inference] [--username USERNAME]
                                        [--backend BACKEND] [--token TOKEN] [--repo-id REPO_ID]
                                        [--push-to-hub] --model MODEL --project-name PROJECT_NAME
                                        [--seed SEED] [--epochs EPOCHS]
                                        [--gradient-accumulation GRADIENT_ACCUMULATION]
                                        [--disable_gradient_checkpointing] [--lr LR] [--log LOG]
                                        [--data-path DATA_PATH] [--train-split TRAIN_SPLIT]
                                        [--valid-split VALID_SPLIT] [--batch-size BATCH_SIZE]
autotrain <command> [<args>] llm: error: argument --model: expected one argument

I have used this before. Why is it not working?

Error Logs

See above.

Additional Information

No response