amrb

6 ms·

1.
by amrb 2y ago
It's a red flag that the 1.2bil model has to fit in gpu memory, happy to be provided wrong when the code drops
2.
by amrb 2y ago
Can't say I mastered the concept either, I'm waiting for the code [0] to be release so I can run some head-to-head tests. [0] https://github.com/Aleph-Alpha/trigrams
3.
by amrb 2y ago
An alternative approache to BPE tokenization https://arxiv.org/abs/2406.19223
4.
by amrb 3y ago
Can check out their project at https://github.com/bigscience-workshop/petals
5.
by amrb 3y ago
Speaking of quantized vectors https://huggingface.co/papers/2309.14717
6.
by amrb 3y ago
Great project and I'm happy to see it expand to more models!
7.
by amrb 3y ago
What's the new reddit to try?
8.
by amrb 3y ago
Anything can end up in logs, then it depends on getting access to hosted splunk via employee creds, for a hypothetical breach.
9.
by amrb 3y ago
There is a salary requirement, as not to under cut local works. Of course if you working over 40 hours a week maybe the company gets it's pound of flesh!
10.
by amrb 3y ago
Good talk on the paper https://www.youtube.com/watch?v=ut5kp56wW_4
11.
by amrb 3y ago
So we just recreated all of the previous SQL injection security issues in LLM's, fun times
12.
by amrb 3y ago
The "one weird trick" to squeeze limes for extra juice
13.
by amrb 3y ago
Group names are 10/10
14.
by amrb 3y ago
Sorry to hear you had this experience, I would say its worth giving another go maybe you can check out the IRC first to check the vibe before committing.
15.
by amrb 3y ago
I appreciate they say they need to learn more about 'exec' when asking GPT4, it also plays well into some of the reading strategies I've seen to get a high-level understanding then read the documentation with more general id
16.
by amrb 3y ago
Sounds cool but I'm not seeing how this discovers IOC's or reverse engineers malware.
17.
by amrb 3y ago
I'd like to see a yearly benchmark for models, could be logic puzzles or a suit of tasks but as it stands there is not good way to measure the ability of models.
18.
by amrb 3y ago
Can some one test fizzbuzz, sounds silly tho a lot of models fail on the combination check in my tests.
19.
by amrb 3y ago
You should look up for hackerspace's in you city, it's a space with tools to do projects.
20.
by amrb 3y ago
We have seen hair works as a Nvidia only tech.. Compatibly could be why the licence theory was rejected.
21.
by amrb 3y ago
What is chatgpt-turbo api pricing like .0001 pre 1k tokens, your paying more in workers salary at the moment.
22.
by amrb 3y ago
A type of battery the small round one.. not the year
23.
by amrb 3y ago
I could see Nvidia licensing the tech to game publishers but they didn't get much uptake, so open source it is.
24.
by amrb 3y ago
feel more likely to rip game assets and build in a supported engine like Unreal.
25.
by amrb 3y ago
The of the widescreen fixes for the Unreal games was the addition of a dll file in the game folder.
26.
by amrb 3y ago
https://en.m.wikipedia.org/wiki/Strange_loop
27.
by amrb 3y ago
Seems the fine tuned models I.e. gpt4all and alpaca are trained as LoRa's. Best advice is jump in and try the demos on hugging space!
28.
by amrb 3y ago
Silly question, but would it be more impactful to use grey or open models to achieve the goals? End of the day the finally model may need to run outside a datacenter so if people don't fine-tune models this could be a limiting factor.
29.
by amrb 3y ago
LoRa has been pretty popular and untill the llama leak was not aware of it, maybe will see something cool out of the open assistant project, we have a lot of English and Spanish prompts and was crazy to see people doing an massive open sour
30.
by amrb 3y ago
Would like to see a yearly benchmark's for models like this!

More ›