
PhD Student at @Berkeley_EECS working on the intersection of large language models and data systems
Joined November 2015
- Tweets73
- Following115
- Followers180
- Likes429
Asim Biswal retweeted
1/6 🚨Releasing Actor-Gatherer framework, a new paradigm for agent context management developed at UC Berkeley for working with multiple agents in parallel over long sessions. Actor-Gatherer decouples knowing from acting: Actor agents perform tasks and Gatherer sets their context
Asim Biswal retweeted
Today, we’re releasing Continual Learning Bench 1.0: the first, realistic benchmark for measuring how AI systems can improve in online settings.
Benchmarks today assume models are stateless. Each example is independent, and once a system finishes a task, it moves on as if nothing happened.
But deployed AI systems should learn from experience. We tested 10+ frontier systems against novel, expert-validated tasks and find there’s still plenty of headroom for learning. (1/n)
NL2SQL agents are already good at writing SQL. But on real large-scale databases, the remaining failures aren't often syntax failures—they come from recurring misconceptions about the data itself.
In Arming Data Agents with Tribal Knowledge, we propose TK-Boost: a bolt-on framework that helps data agents learn reusable, corrective knowledge from past mistakes, then apply it when the same misconception shows up again.
Paper: arxiv.org/abs/2602.13521
Blog: abiswal2001.github.io/TK-Boo…
Code: abiswal2001.github.io/TK-Boo…
🧵👇
(6/n) This work is a collaboration from the Sky Computing Lab on new ways to help AI systems interact more reliably with real data @shagarw21, @SepantaZeighami, @profjoeyg, @adityagp, @alvinkcheung
1
3
113
Asim Biswal retweeted
Some of the most exciting AI apps require LLM reasoning over large datasets at test time.
For these types of NL questions, RAG or Text2SQL + your favorite LLM are simply not enough.
Excited to announce our new leaderboard, from the TAG team at Stanford and Berkeley, to benchmark progress on these difficult types of NL questions over data.
Make a submission & top our leaderboard: huggingface.co/spaces/TAG-Re…
🏆 New Leaderboard Release!
Reasoning models like Deepseek-R1 are excelling on math and science benchmarks, but how well do they handle complex questions over large-scale data?
Table Augmented Generation (TAG) is a unified paradigm that generalizes methods for natural language question answering over data. We are now excited to release the TAG-Leaderboard to evaluate these methods. We are welcoming submissions and want to see how your method does! Joint work from Berkeley and Stanford.
🔗 Leaderboard: huggingface.co/spaces/TAG-Re…
🔗 Project Homepage: tag-research.github.io/homep…
(3/n) We’re actively developing and expanding the benchmark and want to see your submission!
Try it out & submit your model here 👇
huggingface.co/spaces/TAG-Re…
1
1
220
(4/n) This is joint work with the amazing @lianapatel_ @sid_jha1 @AmogKamsetty @LynnLiu41887950 @profjoeyg @matei_zaharia @guestrin
1
208
Asim Biswal retweeted
Instruct-tuned models are getting better at following instructions and ‘reasoning’ every day, but they’re shockingly poor at generating diverse responses. Diversity is crucial to many tasks like synthetic data generation. We tackle this with a new approach, BARE 🐻! (1/n)
Asim Biswal retweeted
We now have a super cool demo of Table Augmented Generation (TAG) that lets you explore important datasets (e.g., US Election Donations).
Try it now! (while we still have LLM inference credits)
The TAG Project combines the language reasoning and world knowledge of LLMs with the computational accuracy and performance of modern database systems.
🧵 Excited to share a demo for our recent work on Table Augmented Generation (TAG)! Ask interesting natural language queries over structured data!
tinyurl.com/tagdemoresearch
Asim Biswal retweeted
This is so cool! Check out this demo from @_asimbiswal for asking questions over large tables, like election contributions🗳️I have so many questions to ask👀
✨And it's powered by Text2LOTUS! NL questions are converted to LOTUS programs for optimized LLM-powered data processing!
🧵 Excited to share a demo for our recent work on Table Augmented Generation (TAG)! Ask interesting natural language queries over structured data!
tinyurl.com/tagdemoresearch
🧵 Excited to share a demo for our recent work on Table Augmented Generation (TAG)! Ask interesting natural language queries over structured data!
tinyurl.com/tagdemoresearch
This demo build on our recent work evaluating Text2SQL and RAG models over realistic natural language questions which found that these methods struggled to answer even 20% of questions correctly.
💡 TAG's secret sauce is effectively combining LLMs and data operators to handle these questions more effectively. More details in our paper! arxiv.org/html/2408.14717v1
A huge thank you to my incredible collaborators , @lianapatel_ , @sid_jha1 , @AmogKamsetty , @LynnLiu41887950 along with faculty mentors @profjoeyg , @matei_zaharia and @guestrin for making this project possible!
Excited to see what interesting questions people come up with!
7
302








