Qwen2 1.5B Summarize

🚀 Qwen2-1.5B-Summarize

Qwen2-1.5B-Instruct finetuned on synthetic data for 2 epochs to perform the summarization task.

🚀 Quick Start

This project is about fine-tuning Qwen2-1.5B-Instruct on custom synthetic data for 2 epochs to perform the summarization task. For more information, visit the GitHub repository.

💻 Usage Examples

Basic Usage

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-1.5B-Instruct")
model = AutoModelForCausalLM.from_pretrained("thepowerfuldeez/Qwen2-1.5B-Summarize", 
                                             bnb_4bit_compute_dtype=torch.bfloat16,
                                             load_in_4bit=True, attn_implementation="flash_attention_2")

text = <YOUR_TEXT>
messages = [
    {"role": "system", "content": "You are helpful AI assistant."},
    {"role": "user", "content": f"Summarize following text: \n{text}"},
]
input_ids = tokenizer.apply_chat_template(messages, return_tensors='pt')
new_tokens = model.generate(input_ids, max_new_tokens=1024)[0][len(input_ids[0]):]
summary = tokenizer.decode(new_tokens, skip_special_tokens=True)

📚 Documentation

Dataset

The train split of the dataset is available here.

Metrics

BERTScore

The BERTScore was calculated using the official implementation with the microsoft/deberta-xlarge-mnli model. 32 inputs were sampled from the test set (longer sentences for summarization), and summaries were generated. The reference summaries were generated by the stronger Qwen2-72B-Instruct model and used as targets for the metric.

📄 License

This project is licensed under the Apache 2.0 license.