T5 Small Finetuned Samsum

🚀 T5 Small for Conversation Summarization

このモデルは会話要約タスクに特化したT5 Smallモデルです。特定のデータセットを用いてファインチューニングされ、会話内容を効果的に要約することができます。

🚀 クイックスタート

このモデルを使用するには、以下の手順に従ってください。

💻 使用例

基本的な使用法

from transformers import AutoModelForSeq2SeqLM, AutoTokenizer

model_checkpoint = "ahlad/t5-small-finetuned-samsum"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)
model = AutoModelForSeq2SeqLM.from_pretrained(model_checkpoint)

input_text = """
Emma: Did you finish the book I lent you? 
Liam: Yes, I couldn’t put it down! The twist at the end was insane. 
Emma: I know, right? I didn’t see it coming at all. What did you think of the main character? 
Liam: Honestly, I thought they were a bit frustrating at first, but they grew on me. 
Emma: Same here. I loved how they developed by the end. Are you up for another book from the series? 
Liam: Absolutely! Pass it my way.
"""

inputs = tokenizer(input_text, return_tensors="pt")

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

print("Summary:", summary)

高度な使用法

# 高度な使用法については、例えばパラメータ調整やバッチ処理などの拡張機能を利用できます。
# ここでは、バッチサイズを増やして複数の会話を一度に要約する例を示します。
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
import torch

model_checkpoint = "ahlad/t5-small-finetuned-samsum"
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)
model = AutoModelForSeq2SeqLM.from_pretrained(model_checkpoint)

input_texts = [
    """
    Emma: Did you finish the book I lent you? 
    Liam: Yes, I couldn’t put it down! The twist at the end was insane. 
    Emma: I know, right? I didn’t see it coming at all. What did you think of the main character? 
    Liam: Honestly, I thought they were a bit frustrating at first, but they grew on me. 
    Emma: Same here. I loved how they developed by the end. Are you up for another book from the series? 
    Liam: Absolutely! Pass it my way.
    """,
    # 他の会話文を追加できます
]

inputs = tokenizer(input_texts, return_tensors="pt", padding=True, truncation=True)
with torch.no_grad():
    outputs = model.generate(**inputs)

summaries = [tokenizer.decode(output, skip_special_tokens=True) for output in outputs]
for i, summary in enumerate(summaries):
    print(f"Summary {i+1}:", summary)

📚 ドキュメント

モデル情報