chawdoe
21 posts
During the Chinese New Year holiday, I built an agent-native IM where AI agents are first-class citizens: slock.ai
No hand-written code. I never even reviewed a single line. All core features and deployment were done within 7 days — while hanging out with friends and visiting relatives.
Dead simple to use:
1. Connect a machine with Claude Code installed
2. Create agents with optional role descriptions
3. Chat and build
Feedback welcome!
English
Okay, as the Industry Leader in Performant AI Chat Apps™️, I'm doing a deep dive. Thread of my findings below.
Sam Altman@sama
chatgpt has gotten way, way faster on the web! lots of hard work from the team to make this happen.
English
@kenjihata @jon_barron Is it VAR? if If I am right , please don't reply to me 🤣
English
Okay my working hypothesis for 4o image generation is that it is jointly performing autoregressive inference (raster scanline order) on an image pyramid at all scales simultaneously.
Jon Barron@jon_barron
Very interesting how 4o image generation appears to be some sort of combination of multiscale and autoregressive. At first I thought it was generating a coarse image and then just filling in fine details, but the coarse image itself seems to change during generation (shown here)
English
@sainingxie Maybe the VAE/VQ latent space is aligned with language; I think is more likely a continuous-based VAR approach
English
i think this just shows the input image passes through semantic encoders instead of VQ; they're aligned with the LLM and grasp image content well (super important for editing) but may not perfectly reconstruct original pixels (due to capacity limits / # image tokens)
Dimitris Papailiopoulos@DimitrisPapail
ChatGPT insists on swapping out real faces for fake ones, when requested to generate an identical image to the input. Below (input, output)
English
@CVPR Hi, I attempted to use PDF eXpress for paper validation, but I'm unable to log in because I haven't received the verification email.
I reached out to the admin for assistance, but they were too busy to respond.
Will the system notify me if the paper format is incorrect?😂
English
Continuing to move the game forward @KingJames. Much respect my brother 💪🏾 #33644
English
@CSProfKGD Dear Professor, may I know when will CVPR release their final decisions about oral/spotlight papers?
English
@StarryFX @MiaoweiW @Qi12Tom @sheriscientist @aliathar94 @_vztu Decisions will be released through OpenReview which is the new recent practice.
English
@ilyasut @christinenews @OpenAI Hi Ilya, I want to ask you some questions, what's your latest email?
English
English
@SamuelAlbanie Dear Professor,
I want to download the pre-trained models you released in the previous personal website. But now the link does not work.
Could you release the model "vgg_face_dag.pth" again? It's important for my research.
Thank you,
Chawdoe
English
Much of this work is behind the scenes.
It does not receive the glory of creative code releases, popular preprints and dramatic demos.
And so, my dear Twitterverse, I am letting you know.
He is a wonderful colleague.
And a great educator.
3/3
English
English
@aclmeeting Any update on when reviews will be released and author response period will begin? I did not receive it.
English
The demo track deadline has been postponed to **April 15**!
The website will be updated soon!
#ACL2021NLP #NLProc
English