Mirrors
/
oai-reverse-proxy
mirror of https://gitgud.io/khanon/oai-reverse-proxy.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

minor gultra fix

This commit is contained in:
nai-degen 2024-08-06 18:46:49 -05:00
parent 9ab1e7d0ce
commit 6c9f302fb9
2 changed files with 7 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/proxy/middleware/request/preprocessors/validate-context-size.ts
View File

@ -81,7 +81,7 @@ export const validateContextSize: RequestPreprocessor = async (req) => {
modelMax = 200000;
} else if (model.match(/^claude-3/)) {
modelMax = 200000;
} else if (model.match(/^gemini-$/)) {
} else if (model.match(/^gemini-/)) {
modelMax = 1024000;
} else if (model.match(/^anthropic\.claude-3/)) {
modelMax = 200000;

10
src/shared/key-management/google-ai/provider.ts
View File

@ -6,7 +6,7 @@ import { getGoogleAIModelFamily, type GoogleAIModelFamily } from "../../models";
import { PaymentRequiredError } from "../../errors";
import { GoogleAIKeyChecker } from "./checker";
// Note that Google AI is not the same as Vertex AI, both are provided by
// Note that Google AI is not the same as Vertex AI, both are provided by
// Google but Vertex is the GCP product for enterprise, while Google API is a
// development/hobbyist product. They use completely different APIs and keys.
// https://ai.google.dev/docs/migrate_to_cloud
@ -103,13 +103,15 @@ export class GoogleAIKeyProvider implements KeyProvider<GoogleAIKey> {
return this.keys.map((k) => Object.freeze({ ...k, key: undefined }));
}
public get(_model: string) {
const availableKeys = this.keys.filter((k) => !k.isDisabled);
public get(model: string) {
const neededFamily = getGoogleAIModelFamily(model);
const availableKeys = this.keys.filter(
(k) => !k.isDisabled && k.modelFamilies.includes(neededFamily)
);
if (availableKeys.length === 0) {
throw new PaymentRequiredError("No Google AI keys available");
}
// (largely copied from the OpenAI provider, without trial key support)
// Select a key, from highest priority to lowest priority:
// 1. Keys which are not rate limited
// a. If all keys were rate limited recently, select the least-recently