Justin Norman
@justintime.ai
š¤ 133
š„ 499
š 244
ML / AI Researcher @ UC Berkeley You can also find me @ justintime.ai
Look, I don't use AI to help me with my review tasks (though I understand the temptation...this is intense, tiring, time expensive, *unpaid!* work) but one thing that really bothers me is when there's a major flaw right out of the gate and you just..have to keep reviewing the paper.
7 days ago
0
0
0
CAN trying to goon Abbey Murphy but if thatās the plan they need to hit harder
8 days ago
0
0
0
Trying to work but Hilary Knight making that impossible by doing GOAT things in Italy
13 days ago
0
1
0
Hahaha Emma Sears was like Iām gettin this one
#uswnt
21 days ago
0
0
0
Good griefā¦phew! Good news in in the District!!
add a skeleton here at some point
27 days ago
0
0
0
I donāt even watch college football but what a game! š
29 days ago
0
0
0
Iāve seen enough: the next PWHL expansion team should be in DC! Absolutely amazing fun
about 1 month ago
0
2
0
Josh Allen what on earth
about 1 month ago
0
1
0
Tonight I archived the project folder for my very first academic paper (and foundation of my dissertation). Itās done. donzo. Fini. Backed up and filed away. End of an era.
about 1 month ago
0
0
0
Inspired, I've been doing this today and it really works! I'm actually *doing* my tasks fully instead of 3,8,12,28 minute interruptions.
add a skeleton here at some point
about 1 month ago
0
0
0
For the past month or so, I've been slowly working through this book by
@sebastianraschka.com
which theoretically and practically builds a GPT model from scratch. Highly recommended! Ironically, I'm writing much more code by hand as a result
about 1 month ago
1
16
2
Hey everyone, I'm happy to announce that the paper I wrote with Alex Hughes and Mike Rivera which explores the lack of truly open, grounded, expert-level hallucination benchmarks for language models has been accepted to ICICT 2026 in Honolulu, HI!
arxiv.org/pdf/2505.17345
about 1 month ago
0
1
0
Hi everyone! Welcome back. I'm so excited to share that the video deepfake detection paper that I co-wrote with Hany Farid has been accepted to the WACV SAFE 2026 workshop!
about 1 month ago
0
1
0
I was totally working (ok limping) my way through DOS 2, but how did I manage to start a Starfield thing two days ago. Watchtower mod is something!
about 2 months ago
0
3
0
All done!
add a skeleton here at some point
about 2 months ago
0
1
0
As my first official act of 2026, I have closed all of my tabs
about 2 months ago
0
0
0
Andā¦Just about finished! just like 2025
add a skeleton here at some point
about 2 months ago
1
1
0
Getting there!
add a skeleton here at some point
about 2 months ago
0
2
1
Also playing a BG3 modded Shadowheart run
#BG3
add a skeleton here at some point
about 2 months ago
0
2
0
Currently playing Dispatch! Highly recommend.
add a skeleton here at some point
about 2 months ago
0
0
0
Another Star Wars set in progress
#lego
add a skeleton here at some point
about 2 months ago
0
1
2
Finally get a chance to read my good friendās new book!
add a skeleton here at some point
about 2 months ago
0
2
0
Started off the break with this awesome TIE Bomber set. I have a lighting kit that Iāll install at some point
add a skeleton here at some point
about 2 months ago
0
1
0
Itās that time of year again when I get to think slowly, learn deeply, and play like I did when I was a kid.
about 2 months ago
0
1
5
Iāll tell you what: whatever team/league that signs Rodman for 2026 and beyond is where my $$$ for sports is going.
3 months ago
0
0
0
Watching
#uswnt
after the NWSL champ letdown as a spirit fan is tough. I need to cheer for Rose, but I donāt like it
3 months ago
1
1
0
Thatās my Washington Spirit going back to the NWSL championship!!!
3 months ago
0
0
0
KINGSBURY!!! Wow!
#nwsl
3 months ago
0
1
0
Finally pulled the trigger and migrated to the
#blacksky
PDS. Took like 5 mins.
5 months ago
0
2
0
Sofia Cantore dear god that was insane
6 months ago
0
0
0
Is Adriana Leon hurt or something?
#uswnt
8 months ago
0
3
0
Do I need to watchā¦US *menās* soccer now? Like, are they good?
#uswnt
8 months ago
0
1
0
Ohh Kay seems like Thompson is just fine that was like a 4.8 40 she just ran
#uswnt
8 months ago
0
1
0
Rose Lavelle is having an absolutely massive match today
#uswnt
8 months ago
0
9
0
Hutton (asked by a reporter about her goal, a part of a 2-0 HT result): āYeah think itās a little sloppy we need to clean it up out thereā Emma Hayes is teaching that championship headspace for
#uswnt
8 months ago
0
1
0
#uswnt
and
#usmnt
both up 2-0 at the same time how about that
8 months ago
0
2
0
Rose out here breaking ankles lmao
#uswnt
8 months ago
0
1
0
Thereās just so much speed on the
#uswnt
roster right now.
8 months ago
0
1
0
I loved the Alters mostly, but the save mechanic is tough if you donāt have uninterrupted time. going to pivot to Expedition 33!
8 months ago
0
0
0
My goodness Alyssa Thompson!
8 months ago
0
0
0
Whoo boy coding on windows is not for the weak
8 months ago
1
0
0
Time for chai and reviewing papers
8 months ago
0
0
0
New Paper! there is little scientific work that attempts to measure the prevalence of language model hallucination in a comprehensive way. We argue that language models should be evaluated using repeatable, open, and domain-contextualized hallucination benchmarking.
arxiv.org/abs/2505.17345
loading . . .
Language models should be subject to repeatable, open, domain-contextualized hallucination benchmarking
Plausible, but inaccurate, tokens in model-generated text are widely believed to be pervasive and problematic for the responsible adoption of language models. Despite this concern, there is little sci...
https://arxiv.org/abs/2505.17345
9 months ago
2
2
0
I have 2 paper deadlines so Iām not really involved but what a time for gaming: AC Shadows, South of Midnight, Atomfall, Clair Obscura 33, and Oblivion remaster (I know, some controversy) Have fun folks!
10 months ago
1
1
0
Checking this one out today, certainly an interesting premise
arxiv.org/abs/2504.03933
loading . . .
Language Models Are Implicitly Continuous
Language is typically modelled with discrete sequences. However, the most successful approaches to language modelling, namely neural networks, are continuous and smooth function approximators. In this...
https://arxiv.org/abs/2504.03933
10 months ago
0
1
0
PIs be like: ācan you meet between 6:14 and 6:23 under the big willow tree? You know the one. Bring all of your results.ā
10 months ago
0
0
0
Iām probably going to do my typical playthrough-pause on AC Shadows early this time. Iām really enjoying South of Midnight
10 months ago
0
1
0
In a short detour from what I probably should be doing, Iām going to check the multimodal world / contrastive learning to extract feature embeddings for audio data
arxiv.org/abs/2211.06687
loading . . .
Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation
Contrastive learning has shown remarkable success in the field of multimodal representation learning. In this paper, we propose a pipeline of contrastive language-audio pretraining to develop an audio...
https://arxiv.org/abs/2211.06687
10 months ago
0
1
0
I read this a couple of weeks ago by Zhou et alāand given the venue and their findings Iām pretty surprised that it didnāt get more attention
www.nature.com/articles/s41...
loading . . .
Larger and more instructable language models become less reliable - Nature
Scaling up and shaping up large language models increased their tendency to provide sensible yet incorrect answers at difficulty levels humans cannot supervise, highlighting the need for a fundam...
https://www.nature.com/articles/s41586-024-07930-y
10 months ago
0
5
2
This has to be the longest sustained period of nice weather in SF Iāve ever experienced
10 months ago
0
0
0
Load more
feeds!
log in