chore: update benchmark using cortex client (cortex-node) (#787)

Author: louis-jan

cortex-js/package.json

@@ -53,7 +53,6 @@
     "decompress": "^4.2.1",
     "js-yaml": "^4.1.0",
     "nest-commander": "^3.13.0",
-    "openai": "^4.50.0",
     "readline": "^1.3.0",
     "reflect-metadata": "^0.2.0",
     "rxjs": "^7.8.1",
@@ -81,6 +80,7 @@
     "@typescript-eslint/eslint-plugin": "^6.0.0",
     "@typescript-eslint/parser": "^6.0.0",
     "bun": "^1.1.15",
+    "cortexso-node": "^0.0.4",
     "cpx": "^1.5.0",
     "eslint": "^8.42.0",
     "eslint-config-prettier": "^9.0.0",

cortex-js/src/domain/models/assistant.interface.ts

@@ -1,5 +1,5 @@
-import { Assistant as OpenAiAssistant } from 'openai/resources/beta/assistants';
-import { AssistantResponseFormatOption as OpenAIAssistantResponseFormatOption } from 'openai/resources/beta/threads/threads';
+import { Assistant as OpenAiAssistant } from 'cortexso-node/resources/beta/assistants';
+import { AssistantResponseFormatOption as OpenAIAssistantResponseFormatOption } from 'cortexso-node/resources/beta/threads/threads';
 
 export interface Assistant extends OpenAiAssistant {
   avatar?: string;

cortex-js/src/domain/models/message.interface.ts

@@ -2,7 +2,7 @@ import {
   Message as OpenAiMessage,
   MessageContent as OpenAiMessageContent,
   TextContentBlock as OpenAiTextContentBlock,
-} from 'openai/resources/beta/threads/messages';
+} from 'cortexso-node/resources/beta/threads/messages';
 
 export interface Message extends OpenAiMessage {}
 

cortex-js/src/domain/models/model.interface.ts

@@ -1,4 +1,4 @@
-import { Model as OpenAiModel } from 'openai/resources/models';
+import { Model as OpenAiModel } from 'cortexso-node/resources/models';
 
 export interface Model
   extends OpenAiModel,

cortex-js/src/domain/models/thread.interface.ts

@@ -1,4 +1,4 @@
-import { Thread as OpenAiThread } from 'openai/resources/beta/threads/threads';
+import { Thread as OpenAiThread } from 'cortexso-node/resources/beta/threads/threads';
 import { Assistant } from './assistant.interface';
 
 export interface ThreadToolResources extends OpenAiThread.ToolResources {}

cortex-js/src/infrastructure/commanders/types/benchmark-config.interface.ts

@@ -1,4 +1,4 @@
-import { ChatCompletionMessageParam } from 'openai/resources';
+import { ChatCompletionMessageParam } from 'cortexso-node/resources';
 
 export interface ApiConfig {
   base_url: string;

cortex-js/src/infrastructure/commanders/usecases/benchmark.cli.usecases.ts

@@ -1,7 +1,7 @@
 import { Injectable } from '@nestjs/common';
 import si from 'systeminformation';
 import fs, { existsSync, mkdirSync, readFileSync, writeFileSync } from 'fs';
-import OpenAI from 'openai';
+import Cortex from 'cortexso-node';
 import { Presets, SingleBar } from 'cli-progress';
 import yaml from 'js-yaml';
 import { FileManagerService } from '@/infrastructure/services/file-manager/file-manager.service';
@@ -27,7 +27,7 @@ export class BenchmarkCliUsecases {
   ) {}
 
   config: BenchmarkConfig;
-  openai?: OpenAI;
+  cortexClient?: Cortex;
   /**
    * Benchmark and analyze the performance of a specific AI model using a variety of system resources
    */
@@ -43,7 +43,7 @@ export class BenchmarkCliUsecases {
 
       const model = params?.model ?? this.config.api.parameters.model;
       // TODO: Using OpenAI client or Cortex client to benchmark?
-      this.openai = new OpenAI({
+      this.cortexClient = new Cortex({
         apiKey: this.config.api.api_key,
         baseURL: this.config.api.base_url,
         timeout: 20 * 1000,
@@ -60,11 +60,7 @@ export class BenchmarkCliUsecases {
         .then(() =>
           this.psUsecases
             .getModels()
-            .then((models) =>
-              models.find(
-                (e) => e.modelId === model,
-              ),
-            ),
+            .then((models) => models.find((e) => e.modelId === model)),
         )
         .then((model) => {
           if (!model)
@@ -147,7 +143,7 @@ export class BenchmarkCliUsecases {
     let firstTokenTime = null;
 
     try {
-      const stream = await this.openai!.chat.completions.create({
+      const stream = await this.cortexClient!.chat.completions.create({
         model: this.config.api.parameters.model,
         messages: this.config.api.parameters.messages,
         max_tokens: this.config.api.parameters.max_tokens,