@element14 , did you ever figure this

jameskraus · 2024-05-13T00:24:19.003Z

@element14 , did you ever figure this out? I'm getting this same `FILE_PARSE_ERROR` with my autotrained finetune

J

jameskrausOP•5/13/24, 12:24 AM

Oh, duh, literally just needed to scroll:
workers-ai

J

jameskrausOP•5/13/24, 12:24 AM

Thanks for following up!

J

jameskrausOP•5/13/24, 12:26 AM

--target_modules q_proj,v_proj

--target_modules q_proj,v_proj

--target_modules q_proj,v_proj

--target_modules q_proj,v_proj

Apparently!

J

jameskrausOP•5/13/24, 1:08 AM

Ah, yeah, that didn't seem to do it

J

jameskrausOP•5/13/24, 1:09 AM

Passed

    "target_modules": "q_proj,v_proj",

    "target_modules": "q_proj,v_proj",

    "target_modules": "q_proj,v_proj",

    "target_modules": "q_proj,v_proj",

But still not accepted :/

J

jameskrausOP•5/13/24, 1:09 AM

FILE_PARSE_ERROR: 'file' should be of valid safetensors type [code: 1000], quiting...

J

jameskrausOP•5/13/24, 1:11 AM

Adapter config:

{
  "alpha_pattern": {},
  "auto_mapping": null,
  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
  "bias": "none",
  "fan_in_fan_out": false,
  "inference_mode": true,
  "init_lora_weights": true,
  "layer_replication": null,
  "layers_pattern": null,
  "layers_to_transform": null,
  "loftq_config": {},
  "lora_alpha": 32,
  "lora_dropout": 0.05,
  "megatron_config": null,
  "megatron_core": "megatron.core",
  "modules_to_save": null,
  "peft_type": "LORA",
  "r": 8,
  "rank_pattern": {},
  "revision": null,
  "target_modules": [
    "q_proj",
    "v_proj"
  ],
  "task_type": "CAUSAL_LM",
  "use_dora": false,
  "use_rslora": false,
  "model_type": "mistral"
}

{
  "alpha_pattern": {},
  "auto_mapping": null,
  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
  "bias": "none",
  "fan_in_fan_out": false,
  "inference_mode": true,
  "init_lora_weights": true,
  "layer_replication": null,
  "layers_pattern": null,
  "layers_to_transform": null,
  "loftq_config": {},
  "lora_alpha": 32,
  "lora_dropout": 0.05,
  "megatron_config": null,
  "megatron_core": "megatron.core",
  "modules_to_save": null,
  "peft_type": "LORA",
  "r": 8,
  "rank_pattern": {},
  "revision": null,
  "target_modules": [
    "q_proj",
    "v_proj"
  ],
  "task_type": "CAUSAL_LM",
  "use_dora": false,
  "use_rslora": false,
  "model_type": "mistral"
}

{
  "alpha_pattern": {},
  "auto_mapping": null,
  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
  "bias": "none",
  "fan_in_fan_out": false,
  "inference_mode": true,
  "init_lora_weights": true,
  "layer_replication": null,
  "layers_pattern": null,
  "layers_to_transform": null,
  "loftq_config": {},
  "lora_alpha": 32,
  "lora_dropout": 0.05,
  "megatron_config": null,
  "megatron_core": "megatron.core",
  "modules_to_save": null,
  "peft_type": "LORA",
  "r": 8,
  "rank_pattern": {},
  "revision": null,
  "target_modules": [
    "q_proj",
    "v_proj"
  ],
  "task_type": "CAUSAL_LM",
  "use_dora": false,
  "use_rslora": false,
  "model_type": "mistral"
}

{
  "alpha_pattern": {},
  "auto_mapping": null,
  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
  "bias": "none",
  "fan_in_fan_out": false,
  "inference_mode": true,
  "init_lora_weights": true,
  "layer_replication": null,
  "layers_pattern": null,
  "layers_to_transform": null,
  "loftq_config": {},
  "lora_alpha": 32,
  "lora_dropout": 0.05,
  "megatron_config": null,
  "megatron_core": "megatron.core",
  "modules_to_save": null,
  "peft_type": "LORA",
  "r": 8,
  "rank_pattern": {},
  "revision": null,
  "target_modules": [
    "q_proj",
    "v_proj"
  ],
  "task_type": "CAUSAL_LM",
  "use_dora": false,
  "use_rslora": false,
  "model_type": "mistral"
}

J

jameskrausOP•5/13/24, 1:14 AM

Currently using huggingface autotrain, since when I use the collab notebook I get this error:

autotrain <command> [<args>] llm: error: argument --quantization/--quantization: invalid choice: 'None' (choose from 'int4', 'int8', None)

autotrain <command> [<args>] llm: error: argument --quantization/--quantization: invalid choice: 'None' (choose from 'int4', 'int8', None)

autotrain <command> [<args>] llm: error: argument --quantization/--quantization: invalid choice: 'None' (choose from 'int4', 'int8', None)

autotrain <command> [<args>] llm: error: argument --quantization/--quantization: invalid choice: 'None' (choose from 'int4', 'int8', None)

(Not sure what's going wrong there, but it doesn't seem like it's compatible with the finetuning instructions

J

jameskrausOP•5/13/24, 1:24 AM

Wow, just removed the

--quantization

--quantization

--quantization

--quantization line in the colab and it appears that unblocked it. I guess None is the default? Hopefully that works

@element14 , did you ever figure this

Similar Threads

Similar Threads

Similar Threads