Blip Base Captioning Ft Hl Narratives

🚀 BLIP-base微調用於敘事圖像字幕生成

BLIP-base是一個基於Transformer架構的模型，本項目在HL Narratives數據集上對其進行微調，以實現高級敘事描述生成，可用於圖像字幕生成等任務。

🚀 快速開始

以下是使用該模型進行圖像字幕生成的示例代碼：

import requests
from PIL import Image
from transformers import BlipProcessor, BlipForConditionalGeneration

processor = BlipProcessor.from_pretrained("blip-base-captioning-ft-hl-narratives")
model = BlipForConditionalGeneration.from_pretrained("blip-base-captioning-ft-hl-narratives").to("cuda")

img_url = 'https://datasets-server.huggingface.co/assets/michelecafagna26/hl/--/default/train/0/image/image.jpg' 
raw_image = Image.open(requests.get(img_url, stream=True).raw).convert('RGB')

inputs = processor(raw_image, return_tensors="pt").to("cuda")
pixel_values = inputs.pixel_values

generated_ids = model.generate(pixel_values=pixel_values, max_length=50,
            do_sample=True,
            top_k=120,
            top_p=0.9,
            early_stopping=True,
            num_return_sequences=1)

processor.batch_decode(generated_ids, skip_special_tokens=True)

>>> "she is holding an umbrella near a lake and is on vacation."

✨ 主要特性

• 微調模型：基於BLIP base模型在HL Narratives數據集上進行微調，專注於高級敘事描述生成。

🔧 技術細節

模型微調 🏋️‍

• 訓練輪數：3個epoch
• 學習率：5e−5
• 優化器：Adam optimizer
• 精度：半精度 (fp16)

測試集指標 🧾

📄 許可證

本項目採用Apache-2.0許可證。

📚 詳細文檔

BibTex和引用信息

@inproceedings{cafagna2023hl,
  title={{HL} {D}ataset: {V}isually-grounded {D}escription of {S}cenes, {A}ctions and
{R}ationales},
  author={Cafagna, Michele and van Deemter, Kees and Gatt, Albert},
  booktitle={Proceedings of the 16th International Natural Language Generation Conference (INLG'23)},
address = {Prague, Czech Republic},
  year={2023}
}