f

.vscode/launch.json

@@ -0,0 +1,22 @@
+{
+    // Use IntelliSense to learn about possible attributes.
+    // Hover to view descriptions of existing attributes.
+    // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+    "version": "0.2.0",
+    "configurations": [
+        {
+            "name": "Python Debugger: BlackJack Game",
+            "type": "debugpy",
+            "request": "launch",
+            "program": "examples/generate_codebase_simple_blackjack.py",
+            "console": "integratedTerminal"
+        },
+        {
+            "name": "Python Debugger: Current File",
+            "type": "debugpy",
+            "request": "launch",
+            "program": "${file}",
+            "console": "integratedTerminal"
+        }
+    ]
+}

examples/experiments/config/config.yaml

@@ -47,5 +47,4 @@ setup:
   api_retry_with_exponential_backoff__exponential_base: 2
   api_retry_with_exponential_backoff__jitter: true
   api_retry_with_exponential_backoff__max_retries: 10
-  api_request_timeout: 60000
   api_stream: False

l2mac/config.py

@@ -30,7 +30,6 @@ class LLMSettingsConfig(BaseModel):
     api_retry_with_exponential_backoff__exponential_base: float = 2
     api_retry_with_exponential_backoff__jitter: bool = True
     api_retry_with_exponential_backoff__max_retries: float = 10
-    api_request_timeout: float = 60000
     api_stream: bool = False
 
 class SetupConfig(BaseModel):

l2mac/llm_providers/general.py

@@ -1,5 +1,6 @@
 from copy import deepcopy
 import re
+from openai import OpenAI, AsyncOpenAI
 import openai
 import asyncio
 from time import sleep, perf_counter
@@ -49,8 +50,8 @@ def get_llm_config(config, logger, name, rate_limiter):
                     'frequency_penalty': config.llm_settings.frequency_penalty,
                     'presence_penalty': config.llm_settings.presence_penalty,
                     'stop': config.llm_settings.stop,
-                    'request_timeout': config.llm_settings.api_request_timeout,
                     'stream': config.llm_settings.api_stream,
+                    'api_key': config.llm.api_key,
                     '_open_ai_rate_limit_requests_per_minute': config.llm_settings.rate_limit_requests_per_minute,
                     '_logger': logger,
                     '_name': name,
@@ -78,6 +79,8 @@ def chat_completion_rl(**kwargs):
     name = kwargs.get('_name', None)
     if kwargs.get('config', None) and kwargs.get('config', None).llm.api_type == "azure":
         use_azure_api = True
+    else:
+        use_azure_api = False
 
     errors: tuple = (openai.RateLimitError,openai.APIError,openai.APITimeoutError,openai.APIConnectionError)
 
@@ -133,7 +136,7 @@ def chat_completion_rl(**kwargs):
         # Raise exceptions for any errors not specified
         except Exception as e:
             raise e
-        
+
 
 async def async_chat_completion_rl_inner(**kwargs):
     logger = kwargs.get('_logger', None)
@@ -155,8 +158,10 @@ async def async_chat_completion_rl_inner(**kwargs):
     # delay_in_seconds = 60.0 / requests_per_minute
     # time.sleep(delay_in_seconds)
 
+    aclient = AsyncOpenAI(api_key=kwargs['api_key'])
     kwargs.pop('_open_ai_rate_limit_requests_per_minute', None)
     kwargs.pop('_logger', None)
+    kwargs.pop('api_key', None)
     kwargs.pop('_name', None)
     kwargs.pop('_rate_limiter', None)
     kwargs.pop('_rate_limiter', None)
@@ -167,9 +172,9 @@ async def async_chat_completion_rl_inner(**kwargs):
 
     if rate_limiter:
         rate_limiter.consume(**kwargs)
-        responses = await openai.ChatCompletion.acreate(**kwargs)
+        responses = await aclient.chat.completions.create(**kwargs)
     else:
-        responses = await openai.ChatCompletion.acreate(**kwargs)
+        responses = await aclient.chat.completions.create(**kwargs)
     response = {}
     chunks = []
     async for chunk in responses:
@@ -217,9 +222,11 @@ def chat_completion_rl_inner(**kwargs):
     # delay_in_seconds = 60.0 / requests_per_minute
     # time.sleep(delay_in_seconds)
 
+    client = OpenAI(api_key=kwargs['api_key'])
     kwargs.pop('_open_ai_rate_limit_requests_per_minute', None)
     kwargs.pop('_logger', None)
     kwargs.pop('_name', None)
+    kwargs.pop('api_key', None)
     kwargs.pop('_rate_limiter', None)
     kwargs.pop('_rate_limiter', None)
     kwargs.pop('stream', None)
@@ -230,9 +237,9 @@ def chat_completion_rl_inner(**kwargs):
 
     if rate_limiter:
         rate_limiter.consume(**kwargs)
-        response = openai.ChatCompletion.create(**kwargs)
+        response = client.chat.completions.create(**kwargs)
     else:
-        response = openai.ChatCompletion.create(**kwargs)
+        response = client.chat.completions.create(**kwargs)
     # if logger:
         # logger.info(f"[{name}][OpenAI API Returned] Elapsed request time: {perf_counter() - t0}s | response: {response}")
     return response