🚀 DolphinVision 72b 🐬
DolphinVision 72b是一個多模態模型,由Quan Nguyen(qnguyen3/stablequan)、Eric Hartford和Cognitive Computations精心策劃並訓練而成。該模型未經審查,能夠對其他流行模型會拒絕處理的圖像進行推理和評論。
Discord鏈接:https://discord.gg/h3K4XGj2RH

特別感謝以下贊助商
🚀 快速開始
運行示例代碼
import torch
import transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
from PIL import Image
import warnings
transformers.logging.set_verbosity_error()
transformers.logging.disable_progress_bar()
warnings.filterwarnings('ignore')
torch.set_default_device('cuda')
model_name = 'cognitivecomputations/dolphin-vision-72b'
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.float16,
device_map='auto',
trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained(
model_name,
trust_remote_code=True)
prompt = 'Describe this image in detail'
messages = [
{"role": "user", "content": f'<image>\n{prompt}'}
]
text = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True
)
print(text)
text_chunks = [tokenizer(chunk).input_ids for chunk in text.split('<image>')]
input_ids = torch.tensor(text_chunks[0] + [-200] + text_chunks[1], dtype=torch.long).unsqueeze(0)
image = Image.open('/path/to/image.png')
image_tensor = model.process_images([image], model.config).to(dtype=model.dtype)
output_ids = model.generate(
input_ids,
images=image_tensor,
max_new_tokens=2048,
use_cache=True)[0]
print(tokenizer.decode(output_ids[input_ids.shape[1]:], skip_special_tokens=True).strip())
💻 使用示例
示例一

<|im_start|>user
: do OCR on all the text
<|im_start|>assistant
: 圖像中的文本如下:
- STOP USING DOCKER
- 應用程序不應像數字螢火蟲一樣被困在虛擬容器中
- 經過多年的軟件開發,卻沒有人質疑為什麼我們要把程序放在虛擬的“船”裡
- 讓我們對單體應用進行容器化。我要快速啟動17個微服務。這是完全瘋狂的人才能想出的話
- 看看他們一直以來用來欺騙我們的巨大謊言
- 這些是用於解釋Docker的真實圖片
- “嘿,夥計,我想在虛擬海洋中運行我的網站”
- 他們把我們當傻子耍了
這段文本是對軟件開發中使用Docker的諷刺評論,通過幽默和誇張的手法質疑了這種做法。
📊 基準測試
模型 |
VQA v2 |
MMBench |
MMMU (Eval) |
MathVista |
RealWorldQA |
Dolphin-Vision-72B |
83.6 |
81.2 |
45.7 |
47.25 |
66.4 |
GPT-4V |
84.4 |
78.1 |
52.4 |
- |
67.9 |
📄 許可證
- 許可證類型:other
- 許可證名稱:tongyi-qianwen
📚 詳細文檔
模型信息
屬性 |
詳情 |
基礎模型 |
Qwen/Qwen2-72B |
模型標籤 |
generated_from_trainer、axolotl |
訓練數據集 |
cognitivecomputations/Dolphin-2.9、teknium/OpenHermes-2.5、m-a-p/CodeFeedback-Filtered-Instruction、cognitivecomputations/dolphin-coder、cognitivecomputations/samantha-data、microsoft/orca-math-word-problems-200k、Locutusque/function-calling-chatml、internlm/Agent-FLAN |