Vach
Vach
I'd like to run the same benchmarks on a different hardware but to have comparable results i need it to be the same code, could you please publish the code...
it is pretty straight forward, anythime i have outline vpn connected i cant comment on youtube i had outline servers hosted in singapore and frankfurt behaviour is the same youtube...
I think you should be doing this so that there is no chance someone is distributing anything else. https://gpt4all.io/models/ggml-gpt4all-j-v1.3-groovy.bin https://huggingface.co/Pi3141/alpaca-native-7B-ggml/resolve/397e872bf4c83f4c642317a5bf65ce84a105786e/ggml-model-q4_0.bin
ggml_new_tensor_impl: not enough space in the context's memory pool (needed 3779301744, available 3745676000) zsh: segmentation fault python3.11 privateGPT.py Whats context memory pool? can i configure it? i actually have a...
Since our best times are in the relm of 2 seconds, imo it is better to run not just 5 times but 10/15 discard outliers and follow the same formula,...
Reading some of the best submissions i think a lot of clever tricks people do never come to play (or even harm the numbers) because C2 will not optimise it...
Not sure whats there, i hope some high level description of the algo, but would be nice to make that link actually reachable..
### Describe the need of your request I have pretty much installed and experimented with most models, some consume significant amount of memory. I wont be surprised if i have...
In most projects one often needs a wrapper type that can carry one message from set of predefined types. In all codecs where this is not supported people end up...
I understand that one can simply run unmodified node projects with this, that's nice but most people would look at graal so they can use node modules out there from...