huac

4 ms·

1.
by huac 2y ago
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, needle in a haystack), but it should be much easier than m
2.
by huac 2y ago
> 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/ , about half of scored submissions score
3.
by huac 2y ago
> Coding assistants based on o1 and Sonnet are pretty great at coding with <50k context, but degrade rapidly beyond that. I had a very similar impression (wrote more in https://hua.substack.com/p/are-longer-contex
4.
Are longer context windows all you need for AI codegen?(hua.substack.com)
1 points by huac 2y ago|0 comments
5.
by huac 2y ago
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 distribution. if it's too high (gibberish like t
6.
by huac 2y ago
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 this a year ago, I estimated the LLM + TTS + other serving cost
7.
by huac 2y ago
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 optimizations) is qui
8.
by huac 2y ago
reminds me a lot of rmarkdown - which allows you to run many languages in a similar fashion https://rmarkdown.rstudio.com/
9.
by huac 2y ago
shouldn't there be more clouds over ocean, as that is where the clouds tend to form?
10.
by huac 2y ago
> 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.
11.
by huac 2y ago
> 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 view but strongly disagree that you need a transcript.
12.
by huac 2y ago
One guess is that the live demo is quantized to run fast on cheaper GPUs, and that degraded the performance a lot.
13.
by huac 2y ago
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 multiple machines (different techniques eg distributed data para
14.
by huac 2y ago
no. only 4 categories can be blocked, while the others cannot be disabled.
15.
by huac 2y ago
His most recent LinkedIn role: Fulfillment Center Associate I, Part Time, Amazon.
16.
by huac 2y ago
> 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 rghit pclae. The rset can be a toatl mses and you c
17.
by huac 2y ago
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 funds in light of the Future of Pensions Act (Wtp). The
18.
by huac 2y ago
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 .
19.
by huac 2y ago
The samples were released a while back: https://google-research.github.io/seanet/stream_vc/
20.
by huac 2y ago
ah nice, did not see this because I searched for `llama-3-v`
21.
Llama-3-V model was plagiarized from MiniCPM(github.com)
21 points by huac 2y ago|6 comments
22.
by huac 2y ago
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 weights were copied from another team's work,
23.
by huac 2y ago
can you elaborate?
24.
by huac 2y ago
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 per month IIRC) - only the data and the systems cha
25.
by huac 2y ago
25% MFU :( maybe because of the P2P nerf?
26.
by huac 2y ago
'sum of the whole is greater than the parts' is a very important line of research to investigate.
27.
by huac 2y ago
the biggest difference is that existing multimodal models (eg GPT-4V and MM1) trained the text model first, and then added in the image component after text training was done ('late fusion'). MM1 learns a projection into the text
28.
by huac 2y ago
Gazelle is (or will be) significantly faster than that.
29.
by huac 2y ago
cat's out of the bag, too early?
30.
by huac 2y ago
related previous discussion https://news.ycombinator.com/item?id=39993626

More ›