metadata

license: apache-2.0
library_name: peft
tags:
  - trl
  - sft
  - generated_from_trainer
datasets:
  - generator
base_model: TheBloke/Mistral-7B-Instruct-v0.1-GPTQ
model-index:
  - name: get_python
    results: []

get_python

This model is a fine-tuned version of TheBloke/Mistral-7B-Instruct-v0.1-GPTQ on the generator dataset. It achieves the following results on the evaluation set:

Loss: 0.5718

Model description

This model can convert a given pseudo-code or algorithm to Python source code.

Intended uses & limitations

This model can be used by reasearchers, students and developers who are struggling to convert algorithms to code.

Training and evaluation data

The model was trained using ananyarn/Algorithm_and_Python_Source_Code.

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
lr_scheduler_warmup_ratio: 0.03
training_steps: 250

Training results

Training Loss	Epoch	Step	Validation Loss
0.8326	0.09	50	0.7046
0.6404	0.18	100	0.6080
0.5771	0.27	150	0.5701
0.5637	0.36	200	0.5662
0.552	0.44	250	0.5718

Framework versions

PEFT 0.8.2
Transformers 4.37.2
Pytorch 2.2.0
Datasets 2.16.1
Tokenizers 0.15.1