robik shrestha retweeted
My research often begins with curiosity-driven questions:
1. Are LLMs Leaking Your Personal Information? (arxiv.org/abs/2205.12628)
2. Are LLMs Really Able to Reason? (arxiv.org/abs/2212.10403)
3. Can LLMs Self-Correct Their Reasoning? (arxiv.org/abs/2310.01798)
4. What is Missing from LLMs? (arxiv.org/abs/2307.02185)
5. Can (Retrieval-Augmented) Encoder-Decoder LMs Perform In-Context Learning? (arxiv.org/abs/2308.07922)
6. Why Do LLMs Fall Short in Providing Truthful Answers? (arxiv.org/abs/2304.10513)
7. Can LMs Be Specific?(arxiv.org/abs/2210.05159) (Note: outdated)
robik shrestha retweeted
Found a kindred spirit at #icml2023 #feelthelearn
robik shrestha retweeted
It’s our mission to better help creatives of all kinds transform the way they imagine, create, and work. We are excited to introduce you to #AdobeFirefly and answer some top-of-mind questions. 🧵adobe.ly/firefly
robik shrestha retweeted
The recordings of our @eccvconf workshop and challenge are now on YouTube.
Check it out if you are interested in OOD generalization in computer vision.
youtube.com/playlist?list=PL…
#ECCV2022
Tomorrow we will host the workshop and challenge on 𝗢𝘂𝘁-𝗼𝗳-𝗗𝗶𝘀𝘁𝗿𝗶𝗯𝘂𝘁𝗶𝗼𝗻 𝗚𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗶𝗻 𝗖𝗼𝗺𝗽𝘂𝘁𝗲𝗿 𝗩𝗶𝘀𝗶𝗼𝗻 @eccvconf #ECCV2022
Tune in to learn about how your models can generalize beyond the training data!
ood-cv.org
robik shrestha retweeted
Can LLMs understand images? We introduce 🔥BLIP-2🔥, a generic and efficient vision-language pre-training strategy that bootstraps from frozen❄️image encoders and frozen❄️LLMs. BLIP-2 outperforms existing SoTAs with only 188M trainable parameters!
Github: github.com/salesforce/LAVIS/…
robik shrestha retweeted
Why haven’t aliens visited our solar system yet?
They looked at the reviews…
Only 1 star.
robik shrestha retweeted
Today, DoNotPay is launching a chatGPT extension that reads the Terms and Conditions, leases and flags anything that is non-standard to all our customers
robik shrestha retweeted
The real AI winner: Microsoft
- $1bn in OpenAI (2019)
- Owns GitHub which released Copilot
- Dalle powers Bing Image Creator
- Valle which produces an AI voice from 3s of input
- GPT in Word, PowerPoint, Outlook etc
- GPT-powered Bing
- $10bn, 49% stake in OpenAI + 75% profit
robik shrestha retweeted
Ready to give your deep models a second life? Introducing model ♻️ recycling (arxiv.org/abs/2212.10445), improving generalization by reusing weights fine-tuned on various vision tasks. Just like you recycle your bottles and cardboards, it's time to start recycling your models too!
robik shrestha retweeted
Weight averaging strategies are super useful in deep learning, and succeed despite the non-linearities in networks' architectures. Our 2 works presented at #NeurIPS2022 analyze how they can help for out-of-distribution classification in computer vision! (1/4)
robik shrestha retweeted
मेन्युमा मम र चाउमिन बाहेक अरु आइटम पनि छन। तर के तपाइ तिनीहरुलाई ट्राइ गर्न र त्यसको मूल्य तिर्न तैयार हुनुहुन्छ ?
robik shrestha retweeted
Going forward, any Twitter handles engaging in impersonation without clearly specifying “parody” will be permanently suspended
robik shrestha retweeted
The importance of choosing your research problem!😈
Two years ago, we pointed out issues in VQA-CP, a benchmark for debiasing (arxiv.org/abs/2005.09241). Dozens of papers were published on the topic (before and after ours). SOTA was claimed and beaten, over and again.⬇️⬇️
robik shrestha retweeted
Want to learn about next generation AI!? Join us for a tutorial on Lifelong Learning Machines at @NeurIPSConf! 😄
With Dhireesha Kudithipudi and @GMvandeVen
sites.google.com/view/neurip…
#NeurIPS2022 #LifelongLearning #ContinualLearning
robik shrestha retweeted
Right one more time.
Happy to announce the release of #StableDiffusion for researchers. Public release soon.
GitHub here: github.com/CompVis/stable-di…
Blogpost (correct link) here: stability.ai/blog/stable-dif…
More on @StabilityAI later when I figure out how to use the internet 🫣
robik shrestha retweeted
In a new interview with Quanta, the computer scientist Christopher Kanan (@chriskanan) speaks with @allisonpwhitten about machine memories, breaking the rules of training neural networks, and whether AI will ever achieve human-level learning. quantamagazine.org/the-compu…



























