MatrixStudio/Instruct-Code-LLaMA-RM-LoRA

CodeLlama-7b-Python-hf

This model is a fine-tuned version of codellama/CodeLlama-7b-Python-hf on the codeforces_python_submissions_rl dataset. It achieves the following results on the evaluation set:

• Loss: 0.3586
• Accuracy: 0.8946

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: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 16
• total_train_batch_size: 32
• total_eval_batch_size: 2
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4458	0.29	500	0.3910	0.8839
0.3026	0.58	1000	0.3721	0.9018
0.3626	0.87	1500	0.3598	0.8946

Framework versions

• PEFT 0.9.0
• Transformers 4.38.2
• Pytorch 2.0.0+cu118
• Datasets 2.18.0
• Tokenizers 0.15.2