UC Berkeley EPIC Lab (@UCBEPIC) header
UC Berkeley EPIC Lab (@UCBEPIC) profile picture

UC Berkeley EPIC Lab

@UCBEPIC

Effective Programming, Interaction, and Computation with Data Lab @UCBerkeley

 Berkeley CA
epic.berkeley.edu
Joined November 2021
  • Tweets88
  • Following22
  • Followers475
UC Berkeley EPIC Lab retweeted
Çağatay Demiralp (@CagatayDemiralp) profile picture
Çağatay Demiralp@CagatayDemiralp20 Apr 2023
This was a great retreat put together by the @UCBEPIC team. As expected, applications of LLMs were central to many talks and discussions. Identifying the core problems invariant to the particularities of the next big LLM is key for research ROI in these applications.
UC Berkeley EPIC Lab (@UCBEPIC) profile pictureUC Berkeley EPIC Lab@UCBEPIC17 Apr 2023
We’re off to the races at the first EPIC data lab retreat in Napa!
2
5
4,701
Fereshte Khani from Microsoft describes how to collaboratively develop NLP models, ensuring alignment and safety
Image from UC Berkeley EPIC Lab (@UCBEPIC)
2
2
3
2,327
It’s hard to robustly test edge cases in a model and make user defined concepts explicit
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
1
422
Can you explore the space using LLMs - but do it in a way that is efficient? How do we find the high error regions?
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
2
327
Alex Bauerle from Sigma Computing tells us about what’s hard when building a spreadsheet for cloud data warehouses
Image from UC Berkeley EPIC Lab (@UCBEPIC)
2
3
8
1,549
A new system they are working on is Humboldt for data discovery. You shouldn’t have to ask experts about what data you should explore!
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
3
420
Naman Jain explores how to summarize data transformation scripts using a template-based approach, informed by LLMs
Image from UC Berkeley EPIC Lab (@UCBEPIC)
2
1
2
863
LLMs by themselves are insufficient for this task - brittle and hard to control
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
2
225
Can you fuse structural understanding of API programs with LLM techniques? Naman provides a way! Parametric templates for the win!
Image from UC Berkeley EPIC Lab (@UCBEPIC)
2
208
Rolando Garcia @rogarcia_sanz describes the next generation of Flor, a tool for rapid iteration during ML training via a live notebook demo!
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
1
3
771
Flor allows users to travel back in time to help debug ML training. You can also inspect and “jump into” another user’s training history. Time travel and shapeshifting!
3
172
Haotian Li describes how to support conversation with data via visualization - why write code when you can just talk to your data!
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
1
6
1,061
Haotian leverages large language models to identify visualization intent (variants of BERT) and prior work on automatically translating visualization intent into actual visualizations (eg Lux).
1
173
Justin Lubin @jplubin embedded himself in a “wet lab” biology group to identify their programming challenges
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
1
3
556
more replies
There is a trade off between easy to understand code (eg one that loops through arrays) and those that are performant (eg one that manipulates arrays in NumPy)
1
1
144
Can we check extensional equality (ie two programs have similar outputs) for constrained domains like biology? So that we can automatically rewrite and make code more performant — component by component?
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
139
Barbara Stanley from Google BQ tells us about user challenges in end to end data analytics/science
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
1
1
483
more replies
More open challenges in helping novice users through the data science workflow - so that one can go from “zero to hero”
Image from UC Berkeley EPIC Lab (@UCBEPIC)
1
1
122
Yet more challenges in Machine Learning - operationalizing, explaining and trusting it.
1
98
Load more