NikkeS/Phi-2-dialogsum-finetuned

Phi-2 Dialogue Summarization Model

Model Description

This is a fine-tuned version of Phi-2, optimized for dialogue summarization. The model is trained on a dataset containing human conversations and their respective summaries, allowing it to generate concise and coherent summaries of dialogue-based texts.

Intended Use

• Summarizing conversations from various sources, including transcripts and chat logs.
• Extracting key points from spoken or written dialogue.
• Assisting in text compression for NLP applications.

Training Details

• Base Model: microsoft/phi-2
• Fine-tuning Method: PEFT (Parameter Efficient Fine-Tuning)
• Dataset: neil-code/dialogsum-test
• Evaluation Metrics: ROUGE scores for summary quality assessment. rouge1: 2.01%, rouge2: -0.29%, rougeL: 1.32%, rougeLsum: 2.53%.

Limitations & Biases

• The model may struggle with highly technical or domain-specific dialogues.
• Potential biases present in the training data could affect summary quality.
• Summarization may sometimes miss nuances in highly informal conversations.

How to Use

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "your-username/phi-2-dialogue-summarization"
tokenizer = AutoTokenizer.from_pretrained(NikkeS/Phi-2-dialogsum-finetuned)
model = AutoModelForCausalLM.from_pretrained(NikkeS/Phi-2-dialogsum-finetuned)

prompt = "Summarize the following conversation:\n\n#Person1#: Hello! How are you?\n#Person2#: I'm good, thanks. How about you?\n\nSummary:"
input_ids = tokenizer(prompt, return_tensors="pt").input_ids
output = model.generate(input_ids, max_length=100)

print(tokenizer.decode(output[0], skip_special_tokens=True))