Viewing profile — huac

that comment refers to the test time inference, i.e. what the model is prompted with, not to what it is trained on. this is, of course, also a tricky problem (esp over long context…

> 32.67% of the successful patches involve cheating as the solutions were directly provided in the issue report or the comments. Looking at the benchmark, https://www.swebench.com/…

> Coding assistants based on o1 and Sonnet are pretty great at coding with I had a very similar impression (wrote more in https://hua.substack.com/p/are-longer-context-windows-all-…

from an AI research perspective -- it's pretty straightforward to mitigate this attack 1. perplexity filtering - small LLM looks at how in-distribution the data is to the LLM's dis…

real-time full duplex like OpenAI GPT-4o is pretty expensive. cascaded approaches (usually about 800ms - 1 second delay) are slower and worse, but very very cheap. when I built thi…

nice work! I wrote a similar library ( https://github.com/stillmatic/gollum/blob/main/packages/vect... ) and similarly found that exact search (w/the same simple heap + SIMD optimi…

reminds me a lot of rmarkdown - which allows you to run many languages in a similar fashion https://rmarkdown.rstudio.com/

shouldn't there be more clouds over ocean, as that is where the clouds tend to form?

> there needs to be a tool/function calling step before a reply I built that almost exactly a year ago :) it was good but not fast enough - hence building the joint model.

> Current AI (even GPT-4o) simply isn't capable enough to do useful stuff. You need to augment it somehow - either modularize it, or add RAG, or similar I am sympathetic to this vi…

One guess is that the live demo is quantized to run fast on cheaper GPUs, and that degraded the performance a lot.

in particular it appears that they only implement data parallel DP - at 1.2B you can fit full copy of model into memory, but larger models require splitting the weights across mult…

no. only 4 categories can be blocked, while the others cannot be disabled.

His most recent LinkedIn role: Fulfillment Center Associate I, Part Time, Amazon.

> Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn't mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the…

I gave the same article to Claude 3.5 Sonnet and the result seems reasonably similar to the author's handwritten summary. ``` This article examines the governance of Dutch pension …

except to the extent that your voice may be part of your image, which is actionable: https://en.wikipedia.org/wiki/Midler_v._Ford_Motor_Co .

The samples were released a while back: https://google-research.github.io/seanet/stream_vc/

ah nice, did not see this because I searched for `llama-3-v`

Last week, a vision-language model made the rounds on Twitter and Hacker News ( https://news.ycombinator.com/item?id=40505099 , made front page). However, the model code and weight…

can you elaborate?

it's probably correct to think of functionally all ML models as being stateless. even something like twitter/fb feed - the models themselves remain the same (usually updated 1-2x p…

25% MFU :( maybe because of the P2P nerf?