|
--- |
|
license: openrail |
|
datasets: |
|
- bigcode/the-stack-dedup |
|
library_name: transformers |
|
tags: |
|
- code_generation |
|
- R programming |
|
- sas |
|
- santacoder |
|
--- |
|
|
|
# Statscoder |
|
|
|
This model is a fine-tuned version of [bigcode/santacoder](https://huggingface.co/bigcode/santacoder) on `R` and `SAS` language repositories in [the stack ](https://huggingface.co/datasets/bigcode/the-stack-dedup) dataset. |
|
|
|
|
|
## Training procedure |
|
|
|
The model was finetuned using the code adapted from [loubnabnl/santacoder-finetuning](https://github.com/loubnabnl/santacoder-finetuning). Adapted to handle multiple subsets of datasets and it is [here](https://github.com/infinitylogesh/santacoder-finetuning). |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 5e-05 |
|
- train_batch_size: 8 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 4 |
|
- optimizer: adafactor |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_steps: 100 |
|
- training_steps: 1600 |
|
- seq_length: 1024 |
|
- no_fp16 |
|
|
