Justin Norman
@justintime.ai
ML / AI Researcher @ UC Berkeley
You can also find me @ justintime.ai
You can also find me @ justintime.ai
That’s my Washington Spirit going back to the NWSL championship!!!
November 15, 2025 at 7:10 PM
That’s my Washington Spirit going back to the NWSL championship!!!
Sofia Cantore dear god that was insane
August 16, 2025 at 12:52 AM
Sofia Cantore dear god that was insane
Hutton (asked by a reporter about her goal, a part of a 2-0 HT result): “Yeah think it’s a little sloppy we need to clean it up out there”
Emma Hayes is teaching that championship headspace for #uswnt
Emma Hayes is teaching that championship headspace for #uswnt
July 3, 2025 at 12:30 AM
Hutton (asked by a reporter about her goal, a part of a 2-0 HT result): “Yeah think it’s a little sloppy we need to clean it up out there”
Emma Hayes is teaching that championship headspace for #uswnt
Emma Hayes is teaching that championship headspace for #uswnt
July 3, 2025 at 12:15 AM
I loved the Alters mostly, but the save mechanic is tough if you don’t have uninterrupted time.
going to pivot to Expedition 33!
going to pivot to Expedition 33!
June 29, 2025 at 4:49 PM
I loved the Alters mostly, but the save mechanic is tough if you don’t have uninterrupted time.
going to pivot to Expedition 33!
going to pivot to Expedition 33!
My goodness Alyssa Thompson!
June 27, 2025 at 2:34 AM
My goodness Alyssa Thompson!
Whoo boy coding on windows is not for the weak
June 20, 2025 at 6:47 PM
Whoo boy coding on windows is not for the weak
Time for chai and reviewing papers
June 16, 2025 at 4:02 PM
Time for chai and reviewing papers
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
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...
arxiv.org
May 28, 2025 at 6:19 AM
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
I have 2 paper deadlines so I’m 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!
Have fun folks!
April 24, 2025 at 3:01 PM
I have 2 paper deadlines so I’m 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!
Have fun folks!
PIs be like: “can you meet between 6:14 and 6:23 under the big willow tree? You know the one. Bring all of your results.”
April 11, 2025 at 3:51 PM
PIs be like: “can you meet between 6:14 and 6:23 under the big willow tree? You know the one. Bring all of your results.”
I’m probably going to do my typical playthrough-pause on AC Shadows early this time. I’m really enjoying South of Midnight
April 10, 2025 at 10:08 PM
I’m probably going to do my typical playthrough-pause on AC Shadows early this time. I’m really enjoying South of Midnight
In a short detour from what I probably should be doing, I’m going to check the multimodal world / contrastive learning to extract feature embeddings for audio data arxiv.org/abs/2211.06687
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...
arxiv.org
April 10, 2025 at 9:52 PM
In a short detour from what I probably should be doing, I’m going to check the multimodal world / contrastive learning to extract feature embeddings for audio data arxiv.org/abs/2211.06687
I read this a couple of weeks ago by Zhou et al—and given the venue and their findings I’m pretty surprised that it didn’t get more attention
www.nature.com/articles/s41...
www.nature.com/articles/s41...
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...
www.nature.com
April 10, 2025 at 9:48 PM
I read this a couple of weeks ago by Zhou et al—and given the venue and their findings I’m pretty surprised that it didn’t get more attention
www.nature.com/articles/s41...
www.nature.com/articles/s41...
This has to be the longest sustained period of nice weather in SF I’ve ever experienced
April 10, 2025 at 8:08 PM
This has to be the longest sustained period of nice weather in SF I’ve ever experienced
The thing about being late in PhD-ing is everything is so dramatic. Get a good result and it’s like: “wow that’s going to be in my DISSERTATION” get a bad one and it’s like “ohnoohnoohno it’s all over I’m never going to graduate I’ve wasted my whole life”
April 10, 2025 at 2:39 PM
The thing about being late in PhD-ing is everything is so dramatic. Get a good result and it’s like: “wow that’s going to be in my DISSERTATION” get a bad one and it’s like “ohnoohnoohno it’s all over I’m never going to graduate I’ve wasted my whole life”