Modelpublic

MalikIbrar/flan-python-expert

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face

0likes22downloads

Model Card

flan-python-expert 🚀

This model is a fine-tuned version of `google/flan-t5-base` on the `codeagent-python` dataset.

It is designed to generate Python code from natural language instructions.

🧠 Model Details

—Base Model: FLAN-T5 Base
—Fine-tuned on: Python code dataset (codeagent-python)
—Task: Text-to-code generation
—Language: English
—Framework: 🤗 Transformers
—Library: adapter-transformers

🏋️ Training

The model was trained using the following setup:

python

from transformers import TrainingArguments

training_args = TrainingArguments(
    output_dir="flan-python-expert",
    evaluation_strategy="epoch",
    learning_rate=2e-6,
    per_device_train_batch_size=1,
    per_device_eval_batch_size=1,
    num_train_epochs=1,
    weight_decay=0.01,
    save_total_limit=2,
    logging_steps=1,
    push_to_hub=False,
)

Trained for 1 epoch

Optimized for low-resource fine-tuning

Training performed using Hugging Face Trainer

Example Usage

python

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

model = AutoModelForSeq2SeqLM.from_pretrained("MalikIbrar/flan-python-expert")
tokenizer = AutoTokenizer.from_pretrained("MalikIbrar/flan-python-expert")

input_text = "Write a Python function to check if a number is prime."
inputs = tokenizer(input_text, return_tensors="pt")

outputs = model.generate(**inputs, max_length=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))