kailasv
@kailasv10
Followers
38
Following
38
Media
1
Statuses
11
Electrical Engineering Ph.D. student @ Stanford
Joined January 2020
Was a pleasure to discuss with Sophia! Check out our discussion on long horizon agents and long context
@kiranvodrahalli shared some cool insights about the training of Gemini 2.5 to play Pokémon, how it handles long context, what self-improvement signs were observed in the model, and other interesting details about the project. Watch the full discussion (below).
0
2
12
Happy to share Michelangelo ( https://t.co/0AgJ7JmToG), a long-context reasoning benchmark which measures performance beyond needle tasks up to arbitrary context lengths and remains challenging for frontier models. Stay tuned for more Michelangelo evals to come!
arxiv.org
We introduce Michelangelo: a minimal, synthetic, and unleaked long-context reasoning evaluation for large language models which is also easy to automatically score. This evaluation is derived via...
3
58
360
We developed TrueImage, an #AI tool guiding patients to take more informative photos for #telemedicine 📷 Our real-time clinical evaluation of TrueImage is now published @JAMADerm. It helped 68% of patients to take better clinical photos in <1 min/patient https://t.co/jqctpnrtw6
1
10
65
To understand if AI tools are helpful in medicine, we have to test them in the real world. In @JAMADerm, we describe developing an AI algorithm to help patients take better telemedicine photos. We also did a prospective pilot with patients! #AAD2023
jamanetwork.com
This quality improvement study assesses whether patient use of an artificial intelligence (AI) algorithm was associated with improved photo quality of skin lesion images taken by patients for...
1
10
43
I’m excited to share a new review on #ML for #cancer diagnosis, prognosis, and treatment in @CellCellPress with @ericwu93, Angela Zhang, @AshAlizadeh, and @james_y_zou. We discuss image and molecular ML for cancer + the regulatory process for cancer ML.
Thrilled to share our new @CellCellPress paper discussing state-of-the-art in #ML for #cancer detection, treatment and monitoring! https://t.co/eHEN1yHodg Topics include ML for: -imaging diagnosis -liquid biopsy -which #ML are FDA-approved/deployed -new omic tech + opportunities
0
6
27
This is our new favorite game: Can you steer generative AI to create a goal image? Stanford scholars @kailasv10 and @james_y_zou launched Art Whisperer, a human-AI collaboration experience. Share your top result on this thread (our best so far is 44): https://t.co/KLBVbX4BCO
2
11
48
Don't do this if you're not ready to get addicted to a new game! Created by @kailasv10 and @james_y_zou! An AI-human collaboration art game - you try to recreate the goal image with prompt engineering: https://t.co/yUStH2rTbM Sadly, my best score so far is 74.
4
6
26
Super excited to release https://t.co/ODdOMPvmnw! 🎮🎨 You collaborate with #AI to create target images and compete with players around🌎. Fun #game + helps us study #humanAI collaboration (3 mins to play). Please share and RT! 🙏 #aiartist #stablediffusion
2
17
82
🎨🎮 Get ready to challenge your artistic skills with ArtWhisperer! 🤖🖼️ Create AI-generated art resembling target images with 3 new challenges released daily. Perfect for AI artists of all levels! Play now at https://t.co/iUzBFrikN7!
#ArtWhisperer #AIart
2
5
15
Excited to share our new @ScienceAdvances paper! We show that 3 skin cancer detection #AI work much worse on dark-skin images and dissect why. We create Diverse Dermatology Images open data to test and reduce bias. Paper https://t.co/aRVMvD3iJc Data+code https://t.co/AqFPvi3vZ4
3
35
158
.@kailasv10 and I describe TrueImage, an algorithm we developed to deal with poor quality photos for telemedicine. We're trying to make good quality clinical images for telemedicine as easy as online mobile check depositing! #psb2021
3
4
21