apunkt/plano
1
0
Fork 0
mirror of https://github.com/katanemo/plano.git synced 2026-05-10 16:22:42 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

use passed in model name in chat completion request (#445)

Browse source
This commit is contained in:
Adil Hafeez 2025-03-21 15:56:17 -07:00 • committed by GitHub
parent bd8004d1ae
commit eb48f3d5bb
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
20 changed files with 364 additions and 89 deletions
Download patch file Download diff file Expand all files Collapse all files

8
crates/common/src/ratelimit.rs
View file

@ -1,7 +1,7 @@
use crate::configuration;
use configuration::{Limit, Ratelimit, TimeUnit};
use governor::{DefaultKeyedRateLimiter, InsufficientCapacity, Quota};
use log::debug;
use log::trace;
use std::fmt::Display;
use std::num::{NonZero, NonZeroU32};
use std::sync::RwLock;
@ -99,9 +99,11 @@ impl RatelimitMap {
selector: Header,
tokens_used: NonZeroU32,
) -> Result<(), Error> {
debug!(
trace!(
"Checking limit for provider={}, with selector={:?}, consuming tokens={:?}",
provider, selector, tokens_used
provider,
selector,
tokens_used
);
let provider_limits = match self.datastore.get(&provider) {