We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
作者大大您好,我在使用DeepSeek-vl-7B多轮对话:第一轮中的第一张图作为one-shot案例,然后第二轮的第二张图需要模型根据第一轮的example进行描述。在这个过程中,大概耗时2-5秒,属于合理范围。如下:
然后我将这个[prompt] * 512,即变成批次处理,大概耗时250多秒,如下:
但是,问题来了,我处理不同的512张图片时,即第二轮需要模型生成的内容中,一个批次512个采用了512张图片,生成速度却异常慢,如下:
请问这是多轮推理效率的问题吗还是其他原因呢?
The text was updated successfully, but these errors were encountered:
user_input_dict = { 'role': 'user', 'content': [ {'type': 'text', 'text': user_input}, {'type': 'image_url', 'image_url': {'url': user_image}} ] } prompts.append(user_input_dict) response = pipe([prompts,....,prompts], gen_config=gen_config) 这是第二轮对话的样例,我的一个批次512个,每个对话中的user_image都不一样时,推理就会异常,耗时很长
Sorry, something went wrong.
还有一个问题就是:如果我在一个对话中使用了两张图片,模型是否可以辨别第一张图、第二张图的位置?
No branches or pull requests
作者大大您好,我在使用DeepSeek-vl-7B多轮对话:第一轮中的第一张图作为one-shot案例,然后第二轮的第二张图需要模型根据第一轮的example进行描述。在这个过程中,大概耗时2-5秒,属于合理范围。如下:
![6bff08173ab0f470fb663d2ea9cf0fc8](https://private-user-images.githubusercontent.com/134385878/345261426-f991e552-0087-422f-8a90-103ba0bf68e5.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MjEwNTg0MzksIm5iZiI6MTcyMTA1ODEzOSwicGF0aCI6Ii8xMzQzODU4NzgvMzQ1MjYxNDI2LWY5OTFlNTUyLTAwODctNDIyZi04YTkwLTEwM2JhMGJmNjhlNS5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjQwNzE1JTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI0MDcxNVQxNTQyMTlaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT02MmU0NzI4MjlmNzAxYmE1NTg1YzU4NDQ3ZGMyZWU4MDgyODkyMDM5Yzk5Zjk5Mzc2ZjI1ZDFkYjlkNmQ3OTUzJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCZhY3Rvcl9pZD0wJmtleV9pZD0wJnJlcG9faWQ9MCJ9.xubY4l6d0RH0eGKhnTnbSLzSKW9q4DBsJxq4FprYgxc)
然后我将这个[prompt] * 512,即变成批次处理,大概耗时250多秒,如下:
![70d58b9aefa85375d96ff2574bdb91d0](https://private-user-images.githubusercontent.com/134385878/345261552-e6dc759e-e17a-4404-9fa1-85d3c1d423dc.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MjEwNTg0MzksIm5iZiI6MTcyMTA1ODEzOSwicGF0aCI6Ii8xMzQzODU4NzgvMzQ1MjYxNTUyLWU2ZGM3NTllLWUxN2EtNDQwNC05ZmExLTg1ZDNjMWQ0MjNkYy5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjQwNzE1JTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI0MDcxNVQxNTQyMTlaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT1lZjQ3MzliMTY3Mjc0YWI3M2VlZWYwNzZkNWE0ZGYyODg3YzNhMGJlOThjZTQxYjVkMzE5OWY2MjY1YzdjZGZmJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCZhY3Rvcl9pZD0wJmtleV9pZD0wJnJlcG9faWQ9MCJ9.SyUCKq_bm5975RqpksSo5S84fBopHadG6zr53NyEaCw)
但是,问题来了,我处理不同的512张图片时,即第二轮需要模型生成的内容中,一个批次512个采用了512张图片,生成速度却异常慢,如下:
![0fa7a5fbb66792d77ff8c0c7b2b989ff](https://private-user-images.githubusercontent.com/134385878/345261786-8b5530ff-7d4e-4b7b-b6a0-a56d0d737f1b.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3MjEwNTg0MzksIm5iZiI6MTcyMTA1ODEzOSwicGF0aCI6Ii8xMzQzODU4NzgvMzQ1MjYxNzg2LThiNTUzMGZmLTdkNGUtNGI3Yi1iNmEwLWE1NmQwZDczN2YxYi5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjQwNzE1JTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI0MDcxNVQxNTQyMTlaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT1hZDgxM2Q5ODE1NmFhMWZhYmQyZTc5MTQ1NzBlZWQxNzZkYzAyMDlkMzRmOTlhOTc5ZTY1MDlhZDhjZjIzZmQ4JlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCZhY3Rvcl9pZD0wJmtleV9pZD0wJnJlcG9faWQ9MCJ9.QX5oIn7-It78D6Du9PL4VLKjWkl282_t7sKs0SFjhtU)
请问这是多轮推理效率的问题吗还是其他原因呢?
The text was updated successfully, but these errors were encountered: