Agent Zero AI framework
api_base fix + litellm update
frdel3b18b8645471
diff --git a/agent.py b/agent.pyindex 1c83e3c46..b9c55145f 100644--- a/agent.py+++ b/agent.py@@ -202,25 +202,13 @@ class AgentContext:agent.handle_critical_exception(e)-@dataclass-class ModelConfig:- provider: models.ModelProvider- name: str- api_base: str = ""- ctx_length: int = 0- limit_requests: int = 0- limit_input: int = 0- limit_output: int = 0- vision: bool = False- kwargs: dict = field(default_factory=dict)-@dataclassclass AgentConfig:- chat_model: ModelConfig- utility_model: ModelConfig- embeddings_model: ModelConfig- browser_model: ModelConfig+ chat_model: models.ModelConfig+ utility_model: models.ModelConfig+ embeddings_model: models.ModelConfig+ browser_model: models.ModelConfigmcp_servers: strprompts_subdir: str = ""memory_subdir: str = ""@@ -582,29 +570,30 @@ class Agent:return models.get_chat_model(self.config.chat_model.provider,self.config.chat_model.name,- **self._get_model_kwargs(self.config.chat_model),+ **self.config.chat_model.build_kwargs(),)def get_utility_model(self):return models.get_chat_model(self.config.utility_model.provider,self.config.utility_model.name,- **self._get_model_kwargs(self.config.utility_model),+ **self.config.utility_model.build_kwargs(),+ )++ def get_browser_model(self):+ return models.get_browser_model(+ self.config.browser_model.provider,+ self.config.browser_model.name,+ **self.config.browser_model.build_kwargs(),)def get_embedding_model(self):return models.get_embedding_model(self.config.embeddings_model.provider,self.config.embeddings_model.name,- **self._get_model_kwargs(self.config.embeddings_model),+ **self.config.embeddings_model.build_kwargs(),)- def _get_model_kwargs(self, model_config: ModelConfig):- kwargs = model_config.kwargs.copy() or {}- if model_config.api_base and "api_base" not in kwargs:- kwargs["api_base"] = model_config.api_base- return kwargs-async def call_utility_model(self,system: str,@@ -670,7 +659,7 @@ class Agent:return response, reasoningasync def rate_limiter(- self, model_config: ModelConfig, input: str, background: bool = False+ self, model_config: models.ModelConfig, input: str, background: bool = False):# rate limiter logwait_log = Nonediff --git a/initialize.py b/initialize.pyindex 8fbd94b31..20ce69bbc 100644--- a/initialize.py+++ b/initialize.py@@ -1,5 +1,5 @@import models-from agent import AgentConfig, ModelConfig+from agent import AgentConfigfrom python.helpers import runtime, settings, deferfrom python.helpers.print_style import PrintStyle@@ -26,7 +26,8 @@ def initialize_agent():return result# chat model from user settings- chat_llm = ModelConfig(+ chat_llm = models.ModelConfig(+ type=models.ModelType.CHAT,provider=models.ModelProvider[current_settings["chat_model_provider"]],name=current_settings["chat_model_name"],api_base=current_settings["chat_model_api_base"],@@ -39,7 +40,8 @@ def initialize_agent():)# utility model from user settings- utility_llm = ModelConfig(+ utility_llm = models.ModelConfig(+ type=models.ModelType.CHAT,provider=models.ModelProvider[current_settings["util_model_provider"]],name=current_settings["util_model_name"],api_base=current_settings["util_model_api_base"],@@ -50,7 +52,8 @@ def initialize_agent():kwargs=_normalize_model_kwargs(current_settings["util_model_kwargs"]),)# embedding model from user settings- embedding_llm = ModelConfig(+ embedding_llm = models.ModelConfig(+ type=models.ModelType.EMBEDDING,provider=models.ModelProvider[current_settings["embed_model_provider"]],name=current_settings["embed_model_name"],api_base=current_settings["embed_model_api_base"],@@ -58,7 +61,8 @@ def initialize_agent():kwargs=_normalize_model_kwargs(current_settings["embed_model_kwargs"]),)# browser model from user settings- browser_llm = ModelConfig(+ browser_llm = models.ModelConfig(+ type=models.ModelType.CHAT,provider=models.ModelProvider[current_settings["browser_model_provider"]],name=current_settings["browser_model_name"],api_base=current_settings["browser_model_api_base"],diff --git a/models.py b/models.pyindex d487c940c..1182f5251 100644--- a/models.py+++ b/models.py@@ -1,3 +1,4 @@+from dataclasses import dataclass, fieldfrom enum import Enumimport loggingimport os@@ -73,6 +74,26 @@ class ModelProvider(Enum):OTHER = "Other OpenAI compatible"+@dataclass+class ModelConfig:+ type: ModelType+ provider: ModelProvider+ name: str+ api_base: str = ""+ ctx_length: int = 0+ limit_requests: int = 0+ limit_input: int = 0+ limit_output: int = 0+ vision: bool = False+ kwargs: dict = field(default_factory=dict)++ def build_kwargs(self):+ kwargs = self.kwargs.copy() or {}+ if self.api_base and "api_base" not in kwargs:+ kwargs["api_base"] = self.api_base+ return kwargs++class ChatChunk(TypedDict):"""Simplified response chunk for chat models."""@@ -233,6 +254,9 @@ class LiteLLMChatWrapper(SimpleChatModel):tokens_callback: Callable[[str, int], Awaitable[None]] | None = None,**kwargs: Any,) -> Tuple[str, str]:++ turn_off_logging()+if not messages:messages = []# construct messages@@ -408,7 +432,9 @@ def _get_litellm_embedding(model_name: str, provider_name: str, **kwargs: Any):def _parse_chunk(chunk: Any) -> ChatChunk:delta = chunk["choices"][0].get("delta", {})- message = chunk["choices"][0].get("model_extra", {}).get("message", {})+ message = chunk["choices"][0].get("message", {}) or chunk["choices"][0].get(+ "model_extra", {}+ ).get("message", {})response_delta = (delta.get("content", "")if isinstance(delta, dict)diff --git a/python/helpers/memory.py b/python/helpers/memory.pyindex ee625b856..f4ee84f39 100644--- a/python/helpers/memory.py+++ b/python/helpers/memory.py@@ -28,7 +28,7 @@ import uuidfrom python.helpers import knowledge_importfrom python.helpers.log import Log, LogItemfrom enum import Enum-from agent import Agent, ModelConfig+from agent import Agentimport modelsimport logging@@ -96,7 +96,7 @@ class Memory:@staticmethoddef initialize(log_item: LogItem | None,- model_config: ModelConfig,+ model_config: models.ModelConfig,memory_subdir: str,in_memory=False,) -> tuple[MyFaiss, bool]:@@ -120,17 +120,10 @@ class Memory:os.makedirs(em_dir, exist_ok=True)store = LocalFileStore(em_dir)- # HOTFIX TODO: unify with agent.py- def _get_model_kwargs(model_config: ModelConfig):- kwargs = model_config.kwargs.copy() or {}- if model_config.api_base and "api_base" not in kwargs:- kwargs["api_base"] = model_config.api_base- return kwargs-embeddings_model = models.get_embedding_model(model_config.provider,model_config.name,- **_get_model_kwargs(model_config),+ **model_config.build_kwargs(),)embeddings_model_id = files.safe_file_name(model_config.provider.name + "_" + model_config.namediff --git a/python/helpers/settings.py b/python/helpers/settings.pyindex 1db6d4e94..1b806de76 100644--- a/python/helpers/settings.py+++ b/python/helpers/settings.py@@ -122,6 +122,7 @@ _settings: Settings | None = Nonedef convert_out(settings: Settings) -> SettingsOutput:from models import ModelProvider+ default_settings = get_default_settings()# main model sectionchat_model_fields: list[SettingsField] = []@@ -381,7 +382,7 @@ def convert_out(settings: Settings) -> SettingsOutput:embed_model_section: SettingsSection = {"id": "embed_model","title": "Embedding Model",- "description": "Settings for the embedding model used by Agent Zero.",+ "description": f"Settings for the embedding model used by Agent Zero.<br><h4>⚠️ No need to change</h4>The default HuggingFace model {default_settings['embed_model_name']} is preloaded and runs locally within the docker container and there's no need to change it unless you have a specific requirements for embedding.","fields": embed_model_fields,"tab": "agent",}@@ -408,6 +409,16 @@ def convert_out(settings: Settings) -> SettingsOutput:})+ browser_model_fields.append(+ {+ "id": "browser_model_api_base",+ "title": "Web Browser model API base URL",+ "description": "API base URL for web browser model. Leave empty for default. Only relevant for Azure, local and custom (other) providers.",+ "type": "text",+ "value": settings["browser_model_api_base"],+ }+ )+browser_model_fields.append({"id": "browser_model_vision",@@ -436,24 +447,6 @@ def convert_out(settings: Settings) -> SettingsOutput:"tab": "agent",}- # # Memory settings section- # memory_fields: list[SettingsField] = []- # memory_fields.append(- # {- # "id": "memory_settings",- # "title": "Memory Settings",- # "description": "<settings for memory>",- # "type": "text",- # "value": "",- # }- # )-- # memory_section: SettingsSection = {- # "id": "memory",- # "title": "Memory Settings",- # "description": "<settings for memory management here>",- # "fields": memory_fields,- # }# basic auth sectionauth_fields: list[SettingsField] = []@@ -829,9 +822,8 @@ def convert_out(settings: Settings) -> SettingsOutput:agent_section,chat_model_section,util_model_section,- embed_model_section,browser_model_section,- # memory_section,+ embed_model_section,stt_section,api_keys_section,auth_section,diff --git a/python/tools/browser_agent.py b/python/tools/browser_agent.pyindex 79be1cedd..6cc40de84 100644--- a/python/tools/browser_agent.py+++ b/python/tools/browser_agent.py@@ -125,11 +125,7 @@ class State:)return result- model = models.get_browser_model(- provider=self.agent.config.browser_model.provider,- name=self.agent.config.browser_model.name,- **self.agent._get_model_kwargs(self.agent.config.browser_model),- )+ model = self.agent.get_browser_model()try:self.use_agent = browser_use.Agent(diff --git a/requirements.txt b/requirements.txtindex a74dfca2a..2a8f994b9 100644--- a/requirements.txt+++ b/requirements.txt@@ -30,7 +30,7 @@ unstructured-client==0.31.0webcolors==24.6.0nest-asyncio==1.6.0crontab==1.0.1-litellm==1.72.4+litellm==1.74markdownify==1.1.0pymupdf==1.25.3pytesseract==0.3.13