Re:Mistral 7B, the most powerful language model for its size...
by yieldcrv on Wednesday September 27, 2023 @11:07AM
(#37675724)
what do you use "the most powerful language model for its size" for?
[ Parent
| on HN ]
Re:Mistral 7B, the most powerful language model for its size...
by ShrigmaMale on Wednesday September 27, 2023 @11:10AM
(#37675778)
Probably a better candidate for local applications? Guessing this was a trial balloon against larger models coming though.
Re:Mistral 7B, the most powerful language model for its size...
by linkpuff on Wednesday September 27, 2023 @11:12AM
(#37675819)
According to their website, it's "Optimal for: low latency, text summarisation, classification, text completion, code completion"
Re:Mistral 7B, the most powerful language model for its size...
by snowram on Wednesday September 27, 2023 @11:11AM
(#37675798)
7B models are small enough to be usable on a smartphone, so a local handheld assistant sounds like a use case.
Re:Mistral 7B, the most powerful language model for its size...
by ComputerGuru on Wednesday September 27, 2023 @11:19AM
(#37675923)
Are they? Unquantized, Llama 2 7b needs over 14GB of GPU (or shared) memory.
Re:Mistral 7B, the most powerful language model for its size...
by polygamous_bat on Wednesday September 27, 2023 @11:27AM
(#37676071)
"Unquantized" is the key word here: with quantization you can get a 4-8x improvement without much performance degradation.
Re:Mistral 7B, the most powerful language model for its size...
by minimaxir on Wednesday September 27, 2023 @12:24PM
(#37677065)
"usable" is not the same as practical.
Even running a quantized and optimized LLM on a smartphone would kill battery life at minimum.
Re:Mistral 7B, the most powerful language model for its size...
by brucethemoose2 on Wednesday September 27, 2023 @12:52PM
(#37677509)
Try MLC-LLM. Its not as bad as you'd think.
In the future(?), they will probably use the AI blocks instead of the GPU, which are very low power.
Re:Mistral 7B, the most powerful language model for its size...
by kirill5pol on Wednesday September 27, 2023 @01:33PM
(#37678136)
One big one is speculative decoding for larger models, the more “accurate”* the smaller model, the more the speed up on the bigger model
* as in matches the token that the larger model would output