Avijit Ghosh
@evijitghosh
Followers
3K
Following
25K
Media
556
Statuses
5K
Technical AI Policy Researcher @huggingface 🤗 . Responsible AI Champion. Leading better AI evals with @evaluatingevals!
Boston, Massachusetts
Joined January 2012
Exactly. Benchmark chasing should not have so much power in the direction of science
0
0
0
Hmmmmm
The hot topic at #ICCV2025 was World Models. They come in different flavors — (interactive) video models, neural simulators, reconstruction models, etc. — but the overarching goal is clear: Generative AI that predict and simulate how the real world works.
0
0
0
"if DNA is truly a language, then we should be able to teach transformers how to write it" absolutely killer blog from @AdeledeHoffer on training a generative DNA model, covering: 🧬k-mer tokenization for genomic data 🧬Custom vocabulary building for DNA 🧬Training a small
6
29
219
🗣️🗣️🗣️
One cannot show that turbojets are safe before actually building turbojets and carefully refining them for reliability. The same goes for AI.
0
0
0
Meaning as the former gets oversaturated and the latter is only getting started, I expect Boston to be the next big AI epicenter 💪
0
0
2
Random off the cuff observation about American AI: LLM folks seem to be concentrated in SF, but AI4Science folks seem to be concentrated in Boston.
1
0
2
🌟 Weekly AI Evaluation Spotlight 🌟 🤖 Did you know malicious actors can exploit trust in AI leaderboards to promote poisoned models in the community? This week's paper 📜"Exploiting Leaderboards for Large-Scale Distribution of Malicious Models" by @iamgroot42 explores this!
1
2
6
Hanging out with Boston participants at the M-Boltz Hackathon, happening parallely in Boston and Darmstadt! The future of AI for Science is open, distributed, and scientists first. Can't wait to see which teams end up on the leaderboard tomorrow!
🚨🧬 Want to build in drug discovery? Join the M-Boltz Hackathon (Oct 20–21, 2025) with @merckgroup & the awesome Boltz team! Tackle challenges in protein, nucleic acid & drug co-folding, scale cutting-edge models, and build the next wave of open science. (+ get to hang with
1
0
2
We're starting a weekly paper spotlight series! Come engage with the posts and let's improve evals together! :) First up: Do Large Language Model Benchmarks Test Reliability?
✨Weekly AI Evaluation Paper Spotlight✨ 🕵️ Is benchmark noise and label errors masking the true fragility of LLMs? 🖇️"Do Large Language Model Benchmarks Test Reliability?" - This paper by @josh_vendrow, @EdwardVendrow @sarameghanbeery @aleks_madry provides insights!
0
0
5
This @COLM_conf Spoltlight work by @vjhofmann @heinemandavidj @nlpnoah is one of the best evaluation works I have read this year. This will be my go-to paper when someone asks what's new in evaluations these days. Some ideas in the paper are so amazing!
3
19
163
Thank goodness HuggingChat is back I deeply missed it
Introducing: HuggingChat Omni 💫 Select the best model for every prompt automatically 🚀 - Automatic model selection for your queries - 115 models available across 15 providers Available now all Hugging Face users. 100% open source.
0
0
0
From a talk I gave last week (being a lil 🌶️ with the title) - most notable AI4Science models and datasets are coming from academic labs. Fund them more! Collab with them more! Do this for free on Hugging Science 🤗
2
0
0
See what we’ve been saying! You need scientists (largely still in Academia) and ML researchers to work together to meaningfully implement AI4Science
An exciting milestone for AI in science: Our C2S-Scale 27B foundation model, built with @Yale and based on Gemma, generated a novel hypothesis about cancer cellular behavior, which scientists experimentally validated in living cells. With more preclinical and clinical tests,
1
0
1
Plots have massively improved yay good job eval people
Introducing Claude Haiku 4.5: our latest small model. Five months ago, Claude Sonnet 4 was state-of-the-art. Today, Haiku 4.5 matches its coding performance at one-third the cost and more than twice the speed.
0
0
2
Trying to start a new hobby and the internet is useless. Maybe AI will finally kill unstructured information retrieval for good and then we will be forced to call or visit friends for help again
0
0
0
We need MOARRRRR 🚀 See: https://t.co/sDvkGwoXRu
atomproject.ai
Reinvigorating AI research in the U.S. by building leading, open models in America
0
0
0
On this topic, it’s important to note that non American companies continue to release frontier open source models at a regular cadence so it is refreshing to see another American org with this mission
Today we're sharing the next phase of Reflection. We're building frontier open intelligence accessible to all. We've assembled an extraordinary AI team, built a frontier LLM training stack, and raised $2 billion. Why Open Intelligence Matters Technological and scientific
1
0
1
I found the article "AI for Scientific Discovery is a Social Problem" from @cgeorgiaw - @evijitghosh, and it blows my mind. Had passed long time before I feel this energy to share knowledge and the wish to contribute in a open science project.
0
1
1
More of such research please! Chatbots are not the future of science, science is
Introducing CellTransformer, a new AI tool developed with @UCSF that makes it easier to explore massive neuroscience datasets and identify important subregions of the brain. 🧵
0
0
3