moving the 'improve' command to turbo mode, with auto_extended=true

pr_agent/algo/__init__.py

@@ -9,6 +9,7 @@ MAX_TOKENS = {
     'gpt-4-0613': 8000,
     'gpt-4-32k': 32000,
     'gpt-4-1106-preview': 128000, # 128K, but may be limited by config.max_model_tokens
+    'gpt-4-0125-preview': 128000,  # 128K, but may be limited by config.max_model_tokens
     'claude-instant-1': 100000,
     'claude-2': 100000,
     'command-nightly': 4096,

pr_agent/algo/pr_processing.py

@@ -11,7 +11,7 @@ from pr_agent.algo.git_patch_processing import convert_to_hunks_with_lines_numbe
 from pr_agent.algo.language_handler import sort_files_by_main_languages
 from pr_agent.algo.file_filter import filter_ignored
 from pr_agent.algo.token_handler import TokenHandler
-from pr_agent.algo.utils import get_max_tokens
+from pr_agent.algo.utils import get_max_tokens, ModelType
 from pr_agent.config_loader import get_settings
 from pr_agent.git_providers.git_provider import FilePatchInfo, GitProvider, EDIT_TYPE
 from pr_agent.log import get_logger
@@ -220,8 +220,8 @@ def pr_generate_compressed_diff(top_langs: list, token_handler: TokenHandler, mo
     return patches, modified_files_list, deleted_files_list, added_files_list
 
 
-async def retry_with_fallback_models(f: Callable):
-    all_models = _get_all_models()
+async def retry_with_fallback_models(f: Callable, model_type: ModelType = ModelType.REGULAR):
+    all_models = _get_all_models(model_type)
     all_deployments = _get_all_deployments(all_models)
     # try each (model, deployment_id) pair until one is successful, otherwise raise exception
     for i, (model, deployment_id) in enumerate(zip(all_models, all_deployments)):
@@ -243,8 +243,11 @@ async def retry_with_fallback_models(f: Callable):
                 raise  # Re-raise the last exception
 
 
-def _get_all_models() -> List[str]:
-    model = get_settings().config.model
+def _get_all_models(model_type: ModelType = ModelType.REGULAR) -> List[str]:
+    if model_type == ModelType.TURBO:
+        model = get_settings().config.model_turbo
+    else:
+        model = get_settings().config.model
     fallback_models = get_settings().config.fallback_models
     if not isinstance(fallback_models, list):
         fallback_models = [m.strip() for m in fallback_models.split(",")]

pr_agent/algo/utils.py

@@ -5,6 +5,7 @@ import json
 import re
 import textwrap
 from datetime import datetime
+from enum import Enum
 from typing import Any, List
 
 import yaml
@@ -15,6 +16,9 @@ from pr_agent.algo.token_handler import get_token_encoder
 from pr_agent.config_loader import get_settings, global_settings
 from pr_agent.log import get_logger
 
+class ModelType(str, Enum):
+    REGULAR = "regular"
+    TURBO = "turbo"
 
 def get_setting(key: str) -> Any:
     try:

pr_agent/settings/configuration.toml

@@ -1,5 +1,6 @@
 [config]
 model="gpt-4" # "gpt-4-0125-preview"
+model_turbo="gpt-4-0125-preview"
 fallback_models=["gpt-3.5-turbo-16k"]
 git_provider="github"
 publish_output=true
@@ -68,17 +69,18 @@ enable_help_text=true
 
 
 [pr_code_suggestions] # /improve #
+max_context_tokens=8000
 num_code_suggestions=4
 summarize = true
 extra_instructions = ""
 rank_suggestions = false
 enable_help_text=true
 # params for '/improve --extended' mode
-auto_extended_mode=false
-num_code_suggestions_per_chunk=8
-rank_extended_suggestions = true
-max_number_of_calls = 5
-final_clip_factor = 0.9
+auto_extended_mode=true
+num_code_suggestions_per_chunk=5
+rank_extended_suggestions = false
+max_number_of_calls = 3
+final_clip_factor = 0.8
 
 [pr_add_docs] # /add_docs #
 extra_instructions = ""

pr_agent/tools/pr_code_suggestions.py

@@ -8,7 +8,7 @@ from pr_agent.algo.ai_handlers.base_ai_handler import BaseAiHandler
 from pr_agent.algo.ai_handlers.litellm_ai_handler import LiteLLMAIHandler
 from pr_agent.algo.pr_processing import get_pr_diff, get_pr_multi_diffs, retry_with_fallback_models
 from pr_agent.algo.token_handler import TokenHandler
-from pr_agent.algo.utils import load_yaml, replace_code_tags
+from pr_agent.algo.utils import load_yaml, replace_code_tags, ModelType
 from pr_agent.config_loader import get_settings
 from pr_agent.git_providers import get_git_provider
 from pr_agent.git_providers.git_provider import get_main_pr_language
@@ -26,6 +26,14 @@ class PRCodeSuggestions:
             self.git_provider.get_languages(), self.git_provider.get_files()
         )
 
+        # limit context specifically for the improve command, which has hard input to parse:
+        if get_settings().pr_code_suggestions.max_context_tokens:
+            MAX_CONTEXT_TOKENS_IMPROVE = get_settings().pr_code_suggestions.max_context_tokens
+            if get_settings().config.max_model_tokens > MAX_CONTEXT_TOKENS_IMPROVE:
+                get_logger().info(f"Setting max_model_tokens to {MAX_CONTEXT_TOKENS_IMPROVE} for PR improve")
+                get_settings().config.max_model_tokens = MAX_CONTEXT_TOKENS_IMPROVE
+
+
         # extended mode
         try:
             self.is_extended = self._get_is_extended(args or [])
@@ -64,10 +72,10 @@ class PRCodeSuggestions:
 
             get_logger().info('Preparing PR code suggestions...')
             if not self.is_extended:
-                await retry_with_fallback_models(self._prepare_prediction)
+                await retry_with_fallback_models(self._prepare_prediction, ModelType.TURBO)
                 data = self._prepare_pr_code_suggestions()
             else:
-                data = await retry_with_fallback_models(self._prepare_prediction_extended)
+                data = await retry_with_fallback_models(self._prepare_prediction_extended, ModelType.TURBO)
 
 
             if (not data) or (not 'code_suggestions' in data):