Re:Using LLaMA with M1 Mac and Python 3.11
by spion on Sunday March 12, 2023 @05:02PM
(#35126245)
Use `--top_p 2 --top_k 40 --repeat_penalty 1.176 --temp 0.7` with llama.cpp
[ Parent
| on HN ]
Re:Using LLaMA with M1 Mac and Python 3.11
by datadeft on Sunday March 12, 2023 @05:33PM
(#35126578)
Not bad with these settings:
./main -m ./models/7B/ggml-model-q4_0.bin \
--top_p 2 --top_k 40 \
--repeat_penalty 1.176 \
--temp 0.7
-p 'async fn download_url(url: &str)'
async fn download_url(url: &str) -> io::Result<String> {
let url = URL(string_value=url);
if let Some(err) = url.verify() {} // nope, just skip the downloading part
else match err == None { // works now
true => Ok(String::from(match url.open("get")?{
|res| res.ok().expect_str(&url)?,
|err: io::Error| Err(io::ErrorKind(uint16_t::MAX as u8))),
false => Err(io::Error
Re:Using LLaMA with M1 Mac and Python 3.11
by yeeeloit on Sunday March 12, 2023 @06:19PM
(#35127034)
What fork are you using?
repeat_penalty is not an option.
It's a new feature :) Pull latest from master.
Re:Using LLaMA with M1 Mac and Python 3.11
by knodi123 on Monday March 13, 2023 @02:18AM
(#35131831)
lol,
./main -m ./models/7B/ggml-model-q4_0.bin \
--top_p 2 --top_k 40 \
--repeat_penalty 1.176 \
--temp 0.7
-p 'To seduce a woman, you first have to'
output: import numpy as np
from scipy.linalg import norm, LinAlgError
np.random.seed(10)
x = -2\*norm(LinAlgError())[0] # error message is too long for command line use
print x [end of text]