README.md

metadata

license: other
tags:
  - generated_from_trainer
base_model: deepseek-ai/deepseek-coder-1.3b-instruct
model-index:
  - name: encoder
    results: []

encoder

This model is a fine-tuned version of deepseek-ai/deepseek-coder-1.3b-instruct on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0670
• Mse: 1.0575
• Rmse: 1.0283
• Mae: 0.9076

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: 5e-05
• train_batch_size: 12
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.01
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Mse	Rmse	Mae
0.0988	0.21	3000	0.0928	1.4795	1.2164	1.1099
0.0793	0.42	6000	0.0867	1.4370	1.1987	1.1076
0.0702	0.63	9000	0.0777	0.7554	0.8691	0.7701
0.0634	0.84	12000	0.0716	1.0950	1.0464	0.9449
0.0563	1.05	15000	0.0686	0.9966	0.9983	0.8899
0.0484	1.26	18000	0.0673	1.0653	1.0321	0.9161
0.0466	1.47	21000	0.0671	1.0877	1.0429	0.9219
0.0462	1.68	24000	0.0670	1.0613	1.0302	0.9090
0.046	1.89	27000	0.0670	1.0575	1.0283	0.9076

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0.dev20230605+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2