John Glover retweeted
I don't wanna live in a world where someone else solves Navier-Stokes better than we do
BREAKING NEWS
The Royal Swedish Academy of Sciences has decided to award the 2024 #NobelPrize in Chemistry with one half to David Baker “for computational protein design” and the other half jointly to Demis Hassabis and John M. Jumper “for protein structure prediction.”
1
172
Our paper on Quality Estimation for summarisation (generating clinical notes from conversations) was presented at Interspeech during the week.
tl;dr: metric agnostic QE. Train another generation model, sample from it lots of times, and use the generations as (pseudo)references👇
2
176
Reference-Free Estimation of the Quality of Clinical Notes Generated from Doctor-Patient Conversations: isca-archive.org/interspeech…
1
64
Not every day you get to chat to @geoffreyhinton about consciousness at the bar :)
Fun event at UCD
4
275
Seeing the usual worst-people-in-tech competitors relentlessly chipping away at one of the few good things to come out of the pandemic - the normalisation of remote work - is depressing, and yet entirely unsurprising.
3
178
Another excellent Ted Chiang piece:
newyorker.com/science/annals…
Highlights of some of the machine learning research going on in our wider group at 3M Health: insideangle.3m.com/his/blog-…
126
John Glover retweeted
🖼️ The Big Picture Workshop @ EMNLP 2023
We invite you to share the "big picture", the larger story to which a paper is connected.
bigpictureworkshop.com/
Happy to say that we got mentioned in the #gemworkshop22 outstanding papers list, taking the runner up spot in the "Outstanding Evaluation Insight" category.
Really enjoyed workshop yesterday, thanks to all of the organisers.
With all of the buzz around ChatGPT, and the potential for LLMs to generate false information, seems like a good time to discuss our new paper on detecting factual inconsistencies in summaries: arxiv.org/abs/2211.16853
👇
John Glover retweeted
Two papers to be presented at #EMNLP2022 workshops👇
With all of the buzz around ChatGPT, and the potential for LLMs to generate false information, seems like a good time to discuss our new paper on detecting factual inconsistencies in summaries: arxiv.org/abs/2211.16853
👇
In general, you will still need to carefully tune some parameters/settings for your downstream task of interest, we don't see a single method that works best across all of the datasets that we evaluate on (even though all contain very similar types of content).
1










