rhdunn

30 ms·

361.
by rhdunn 3y ago
Pay is not the only thing regarding maintainership. Time is another factor. It takes time to maintain software, improve the codebase, add features, etc. Then there are the other tasks such as answering questions, reviewing PRs, triaging bug
362.
by rhdunn 3y ago
As a personal anectote, I find anything that moves in my periferal vision to be distracting. Therefore, if there are any animations that occur without me initiating them on a second monitor, I find those distracting. That includes animated
363.
by rhdunn 3y ago
NIMSS -- Not In My Solar System
364.
by rhdunn 3y ago
Probably not many. However, 4090s would be a different situation. There are plenty of guides on running LLMs, stable diffusion, etc. on local hardware. The H100s would be for businesses looking to get into this space.
365.
by rhdunn 3y ago
I'm using pyenv to manage different projects on my machine. I use --system-site-packages to share libraries like torch, and some others -- especially to ensure those libraries are built with the same cuda version, avoid some issues whe
366.
by rhdunn 3y ago
Where KYC means Know Your Customer.
367.
by rhdunn 3y ago
My layperson understanding is that GR specifies that energy/mass-momemtum influences space-time curvature which then produces gravity (as particles travelling along straight lines in the curved space). That would imply that fermions ar
368.
by rhdunn 3y ago
This is why I tend to lean toward the approach of only using mocks, stubs, etc. when absolutely needed. That tends to be at the places where the code is interacting with external components like databases or web services than with component
369.
by rhdunn 3y ago
There are several UI/UX things I hate about reddit that a competitor could easily fix: 1. Clicking on the side of the page -- on the background -- navigates to the subreddit main page. This makes it easy to navigate away when e.g. clic
370.
by rhdunn 3y ago
Just like how Leibniz and Newton independantly formulated calculus, building on the work of others.
371.
by rhdunn 3y ago
It's an appendix (supplementary material) describing Hidden Markov Models. It is web only presumably because a) it is not specific to NLP, and b) keeping it out reduces the print size/cost.
372.
by rhdunn 3y ago
It depends on what you are trying to do. LLMs are powerful, but can be difficult to work with w.r.t. doing post processing or adding markup/annotations in the text. For example, asking it to label terms in a sentence it can produce var
373.
by rhdunn 3y ago
I find the interplay between audio codecs, speech synthesis, and speech recognition fascinating. Advancements in one usually results in advancements in the others.
374.
by rhdunn 3y ago
Interesting. So that would mean that you would still need a 40 or 80 GB card to run the larger models (30B LLM, 70B LLM, 8x7B LLM) and perform training of them. Or would it be possible to split the model layers between the cards like you ca
375.
by rhdunn 3y ago
The A100 is comparable to the 3090 but with more memory. The H100 is the one comparable to the 4090. The advantage of these is the access to the larger memory. And they are able to be linked together such that they all share the same memory
376.
by rhdunn 3y ago
Then specify that in your prompt. "... in the style of ..." or "... in a ... setting". The point is that those modifications should be reliable, so if you want a viking man/woman or an asian/african/greek
377.
by rhdunn 3y ago
You can geolocate a user/request based on the IP address. This is common in servers. It allows a server to know which country a request is coming from. I suspect that this is what OpenAI are doing, as Bing Translate and Google Translat
378.
by rhdunn 3y ago
I use vim to edit git commit messages and other git related tasks as I'm most comfortable/used to working with the git cli (muscle memory, etc.). I also use either vim or nano when viewing/editing files on the command line (e
379.
by rhdunn 3y ago
That doesn't factor in line delivery. You can have the words say/mean one thing (e.g. "I'm fine.") and the delivery say/mean another (defensive, distraught, etc.). It also does not account for where stresses, e
380.
by rhdunn 3y ago
Cash is useful in cases where your phone/card isn't working. E.g. if your phone has run out of battery, you are in a location that has a poor signal, your account has been locked due to a suspicious/unusual transaction (even
381.
by rhdunn 3y ago
We need some Gitea sent to the GitLab! :) Make people aware that there are more options for self-hosted git.
382.
by rhdunn 3y ago
That's interesting. It also looks like koboldcpp works better with long interactions, as it only processes changed tokens. I'm using llama.cpp with text-generation-webui and its OpenAI compatible API. I'll have to look to see
383.
by rhdunn 3y ago
That was what I was referring to with the 32/64 GB systems.
384.
by rhdunn 3y ago
If you want to run Mixtral 8x7B locally you can use llama.cpp (including with any of the supporting libraries/interfaces such as text-generation-webui) with https://huggingface.co/TheBloke/Nous-Hermes-2-Mixtral-8x7
385.
by rhdunn 3y ago
Is there any work/progress on a NN/trained model based on Internet Phonetic Alphabet (IPA) transcriptions? I.e. to be able to create an IPA transcription and convert it back to sound. That approach would be useful for things like
386.
by rhdunn 3y ago
The zip files contain the complete data in various formats. For the WNDB (WordNet DataBase) format, you could use the WordNet viewer or any WordNet compatible software. This will not be able to view all the Open English WordNet data, as it
387.
by rhdunn 3y ago
And Open English WordNet ( https://en-word.net/ ) which builds on WordNet 3.0 and 3.1, improving the core data both with new words and with fixes to the hierarchy.
388.
by rhdunn 3y ago
WordNet has antonym and other semantic relationship information, so it should be possible to add that to GraphWords.
389.
by rhdunn 3y ago
Semantic. Specifically, showing the synonym relationships in the WordNet database. Those are manually constructed.
390.
by rhdunn 3y ago
So that would mean articles from the 1920s, provided that the authors of those articles have been dead for 70 years, or longer in some other countries.

More ›