Sam-Shin/python3-text

OLMo Code Python3 Text-Only Model

This is a LoRA adapter fine-tuned on the OLMo-1B model for Python 3 code generation tasks.

Model Details

• Base Model: allenai/OLMo-1B-hf
• Model Type: LoRA Adapter
• Task: Causal Language Modeling for Python 3 code
• Language: Python 3
• License: MIT
• Fine-tuned by: dipikakhullar

Model Description

This model is a LoRA adapter that has been fine-tuned on Python 3 code data. It extends the capabilities of the base OLMo-1B model specifically for Python code generation tasks.

LoRA Configuration

• LoRA Type: LORA
• LoRA Alpha: 16
• LoRA Dropout: 0.05
• LoRA Rank (r): 8
• Target Modules: down_proj, q_proj, v_proj, up_proj, k_proj, gate_proj, o_proj
• Task Type: CAUSAL_LM

Uses

Direct Use

This model is intended for Python 3 code generation tasks. It can be used to:

• Generate Python code completions
• Assist with code writing
• Provide code suggestions

Downstream Use

The model can be further fine-tuned for specific Python programming tasks or integrated into code generation applications.

Out-of-Scope Use

This model is specifically designed for Python 3 code generation and may not perform well for:

• Other programming languages
• Natural language tasks
• Non-code related tasks

How to Get Started with the Model

from peft import PeftModel, PeftConfig
from transformers import AutoModelForCausalLM, AutoTokenizer

# Load the base model and tokenizer
base_model = AutoModelForCausalLM.from_pretrained("allenai/OLMo-1B-hf")
tokenizer = AutoTokenizer.from_pretrained("allenai/OLMo-1B-hf")

# Load the LoRA adapter
model = PeftModel.from_pretrained(base_model, "dipikakhullar/olmo-code-python3-text-only")

# Example usage
prompt = "def fibonacci(n):"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=100, temperature=0.7)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Training Details

Training Data

The model was fine-tuned on cleaned Python 3 code data specifically prepared for language model training.

Training Procedure

• Base Model: allenai/OLMo-1B-hf
• Fine-tuning Method: LoRA (Low-Rank Adaptation)
• Checkpoint: checkpoint-6000

Model Card Contact

• Author: dipikakhullar
• Repository: https://huggingface.co/dipikakhullar/olmo-code-python3-text-only

Framework versions

• PEFT 0.7.1
• Transformers