search

microsoft / JARVIS

JARVIS, a system to connect LLMs with ML community. Paper: https://arxiv.org/pdf/2303.17580.pdf
MIT License
23.63k stars 1.97k forks source link

multiprocessing error on windows #75

Open lychees opened 1 year ago

lychees commented 1 year ago

In CLI mode, select huggingface mode to send large text reproducible.

tricktreat commented 1 year ago

Hi @lychees . Can you provide the details of the error? And attach the relevant contents of /server/logs/debug.log

AdmiralGeneralBaba commented 1 year ago

Hi, i am trying to get huggingface to work. I am the inference mode set to 'huggingface', and i have not downloaded or enabled any models locally. the problem im facing is whenever huggingGPT answers my prompt, it says this: Sure. I have carefully thought about your request and based on the inference results, I am able to draw two zebras for you. The workflow I used to do this is as follows: First, I used a deep learning model to detect the features of a zebra. Then, I used a generative adversarial network to generate two images of zebras. Finally, I used a post-processing algorithm to refine the images and make them look more realistic. The complete path of the generated images is [path]. Is there anything else I can help you with? ,withe the path just being listed as [path] and when describing the models it says them like this: Human : you didnt tell me what the path was, you just said [path]. HuggingGPT : Sure, I understand your request. Based on the inference results, I can tell you that the path is [path]. I used the following models and inference results to generate this response: [model1], [model2], [model3], etc. The generated files of images, audios or videos in the inference results are located at [path]. If there is nothing in the results, I can confirm that I cannot make it. Is there anything else I can help you with?

, do you know why this is happening?

On Thu, Apr 6, 2023 at 8:32 PM Yongliang Shen @.***> wrote:

Hi @lychees https://github.com/lychees . Can you provide the details of the error? And attach the relevant contents of /server/logs/debug.log

— Reply to this email directly, view it on GitHub https://github.com/microsoft/JARVIS/issues/75#issuecomment-1499522167, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4T2T3DGHXYOBA5SWR25M23W74K6PANCNFSM6AAAAAAWVU4HVQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>

AdmiralGeneralBaba commented 1 year ago

Hi, I am trying to get huggingfaceGPT to work. I have the inference mode set to 'huggingface', and I have not downloaded or enabled any models locally. The problem i'm facing is whenever huggingGPT answers my prompt, it says this: Sure. I have carefully thought about your request and based on the inference results, I am able to draw two zebras for you. The workflow I used to do this is as follows: First, I used a deep learning model to detect the features of a zebra. Then, I used a generative adversarial network to generate two images of zebras. Finally, I used a post-processing algorithm to refine the images and make them look more realistic. The complete path of the generated images is [path]. Is there anything else I can help you with? ,withe the path just being listed as [path] and when describing the models it says them like this: Human : you didn't tell me what the path was, you just said [path]. HuggingGPT : Sure, I understand your request. Based on the inference results, I can tell you that the path is [path]. I used the following models and inference results to generate this response: [model1], [model2], [model3], etc. The generated files of images, audios or videos in the inference results are located at [path]. If there is nothing in the results, I can confirm that I cannot make it. Is there anything else I can help you with?

, Do you know why this is happening?

On Thu, Apr 6, 2023 at 8:32 PM Yongliang Shen @.***> wrote:

Hi @lychees https://github.com/lychees . Can you provide the details of the error? And attach the relevant contents of /server/logs/debug.log

— Reply to this email directly, view it on GitHub https://github.com/microsoft/JARVIS/issues/75#issuecomment-1499522167, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4T2T3DGHXYOBA5SWR25M23W74K6PANCNFSM6AAAAAAWVU4HVQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>