|
--- |
|
license: apache-2.0 |
|
datasets: |
|
- ai-aerospace/ams_data_train_mistral_v0.1_100 |
|
base_model: mistralai/Mistral-7B-Instruct-v0.1 |
|
--- |
|
First model fine tune trained from https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1. |
|
Code to create this here: https://colab.research.google.com/drive/1Wsi7q1sBJlXrVZAbxhMRZuKnhSFeU9mu?usp=sharing |
|
|
|
Parameters used for fine tuning: |
|
`model_params={ |
|
"project_name": project_name, |
|
"model_name": model_name, |
|
"repo_id": username+'/'+repo_name, |
|
"block_size": block_size, |
|
"model_max_length": max_token_length, |
|
"logging_steps": -1, |
|
"evaluation_strategy": "epoch", |
|
"save_total_limit": 1, |
|
"save_strategy": "epoch", |
|
"mixed_precision": "fp16", |
|
"lr": 0.00003, |
|
"epochs": 3, |
|
"batch_size": 1, |
|
"warmup_ratio": 0.1, |
|
"gradient_accumulation": 1, |
|
"optimizer": "adamw_torch", |
|
"scheduler": "linear", |
|
"weight_decay": 0, |
|
"max_grad_norm": 1, |
|
"seed": 42, |
|
"quantization": "int4", |
|
"lora_r": 16, |
|
"lora_alpha": 32, |
|
"lora_dropout": 0.05 |
|
}` |
