-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
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
Computer requirements #2
Comments
Hi, since our model was aligned with Vicuna 13B. The configuration requires approximately 23GB of GPU memory, and we recommend using an RTX 3090 at least. @pixeli99 will provide more details later. |
Hi, Thank you very much for your attention. We are about to update a new version to gain stronger dialogue ability and suppress the shortcomings of the model's "fantasy". We will improve readme when the new version is updated. In addition, a current version of the web demo has already been launched. You can experience it (although the link to this demo may expire, we will update it in a timely manner) |
When I clicked, the demo had already expired, but the new README looks great! Everything is greatly explained. Happy to schedule a call in the near future to brainstorm how to overcome the 'hallucinations' |
Thanks Jon. We are trying several methods to get a better understanding of
the video and also help GPT understand the video caption.
Will have a call in the summer when we have a more clean solution.
Thank you so much for your very kind support~
BW,
Xiaowei
…On Fri, 19 May 2023 at 09:54, Jonathan Cardoso-Silva < ***@***.***> wrote:
When I clicked, the demo had already expired, but the new README looks
great! Everything greatly explained.
Happy to schedule a call in the near future to brainstorm how to overcome
the 'hallucinations'
—
Reply to this email directly, view it on GitHub
<#2 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AU4VDIQCZEJ2NDENEKSF4X3XG4YLTANCNFSM6AAAAAAX3SEHRA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
--
高晓伟
Xiaowei Gao
Ph.D. Student of UCL
|
Thank you very much for your feedback. We're currently in the process of further developing this project and plan to update the demo link at a later time, due to our limited GPU resources.
We're eager to discuss potential solutions for the 'hallucinations' issue and would greatly value your insights.
By the way, I noticed that you're scheduled to give a talk at ICL&LSE on the 24th and 25th. If possible, we could have a brief discussion or catch up there.
…________________________________
From: Jonathan Cardoso-Silva ***@***.***>
Sent: 19 May 2023 9:54
To: pixeli99/DSify ***@***.***>
Cc: Ming Li ***@***.***>; Comment ***@***.***>
Subject: Re: [pixeli99/DSify] Computer requirements (Issue #2)
When I clicked, the demo had already expired, but the new README looks great! Everything greatly explained.
Happy to schedule a call in the near future to brainstorm how to overcome the 'hallucinations'
—
Reply to this email directly, view it on GitHub<#2 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AGXJVST74LZA65XDDKFG6ILXG4YLTANCNFSM6AAAAAAX3SEHRA>.
You are receiving this because you commented.Message ID: ***@***.***>
|
Hi @mmingli, definitely, let's chat at the event. Feel free to 'chase' me there! |
Great! Looking forward to catching up with you at the event.
…________________________________
From: Jonathan Cardoso-Silva ***@***.***>
Sent: 19 May 2023 12:01
To: pixeli99/DSify ***@***.***>
Cc: Ming Li ***@***.***>; Mention ***@***.***>
Subject: Re: [pixeli99/DSify] Computer requirements (Issue #2)
Hi @mmingli<https://github.com/mmingli>, definitely, let's chat at the event. Feel free to 'chase' me there!
—
Reply to this email directly, view it on GitHub<#2 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AGXJVSWJ5GRKQ2WS3VKCEF3XG5HGZANCNFSM6AAAAAAX3SEHRA>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
Hey guys, @pixeli99 and @UCLWilson, awesome work you are doing!
I was trying to replicate the demo on Mac Studio, but I see that MiniGPT4 relies on CUDA so it won't be super easy for me to replicate (I'll try later).
Could you add the setup you are using to run the current demos? Which OS, how much RAM and GPU you are using at the moment, etc.?
The text was updated successfully, but these errors were encountered: