finetune_starcoder2_with_R_data

This model is a fine-tuned version of bigcode/starcoder2-3b, adapted and fine-tuned specifically for generating R programming code.

Model description

This model is a specialized version of the bigcode/starcoder2-3b architecture fine-tuned on a subset of the Stack dataset, focusing solely on R programming language data. The fine-tuning process utilized the PEFT (Parameter Efficient Fine Tuning) method and included loading the model with 4-bit quantization using the LoRA library. It's tailored for generating R programming code, offering optimized performance for tasks within this domain.

Intended uses & limitations

Tailored for R programming tasks, this model is optimized for generating code snippets, functions, or scripts in the R language. Its limitations may include its applicability solely within the domain of R programming and potential constraints related to the size and diversity of the training data.

Training and evaluation data

The model was trained and evaluated on a subset of the bigcode/Stack dataset containing R programming language data.

Training procedure

Fine-tuning was performed using the PEFT (Parameter Efficient Fine Tuning) method over 1000 epochs on the R dataset. Additionally, the model was loaded with 4-bit quantization using the LoRA library to optimize memory usage and inference speed, enhancing its efficiency for generating R code.

Training hyperparameters

The following hyperparameters were used during training: - learning_rate: 0.0002 - train_batch_size: 2 - eval_batch_size: 16 - seed: 0 - gradient_accumulation_steps: 4 - total_train_batch_size: 8 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: cosine - lr_scheduler_warmup_steps: 100 - training_steps: 1000 - mixed_precision_training: Native AMP

Framework versions

• PEFT 0.8.2
• Transformers 4.40.0.dev0
• Pytorch 2.1.2
• Datasets 2.16.1
• Tokenizers 0.15.2