metadata

base_model: shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat
library_name: peft
license: other
tags:
  - llama-factory
  - lora
  - generated_from_trainer
model-index:
  - name: Mistral-7B-v0.3-Chinese-Chat
    results: []

Mistral-7B-v0.3-Chinese-Chat

This model is a fine-tuned version of shenzhi-wang/Mistral-7B-v0.3-Chinese-Chat on the alpaca_mac dataset. It achieves the following results on the evaluation set:

Loss: 2.0317

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 16
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 128
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.1
num_epochs: 6.0

Training results

Training Loss	Epoch	Step	Validation Loss
1.6196	0.9964	35	1.5316
1.3654	1.9929	70	1.4583
1.0796	2.9893	105	1.5236
0.7613	3.9858	140	1.7009
0.5639	4.9822	175	1.8934
0.4648	5.9786	210	2.0317

Framework versions

PEFT 0.11.1
Transformers 4.43.3
Pytorch 2.4.0+cu121
Datasets 2.19.1
Tokenizers 0.19.1