Automate browser based workflows with AI
Python aiapiautomationbrowserbrowser-automationcomputergptllmplaywrightpowerautomatepuppeteerpythonrpaseleniumvisionworkflow
Update max output tokens to 16K (#1110)
Shuchang Zhengbf06fcfeb759
diff --git a/skyvern/forge/sdk/api/llm/api_handler_factory.py b/skyvern/forge/sdk/api/llm/api_handler_factory.pyindex d73519b55..55b458fa4 100644--- a/skyvern/forge/sdk/api/llm/api_handler_factory.py+++ b/skyvern/forge/sdk/api/llm/api_handler_factory.py@@ -74,7 +74,7 @@ class LLMAPIHandlerFactory:The response from the LLM router."""if parameters is None:- parameters = LLMAPIHandlerFactory.get_api_parameters()+ parameters = LLMAPIHandlerFactory.get_api_parameters(llm_config)if step:await app.ARTIFACT_MANAGER.create_artifact(@@ -168,7 +168,7 @@ class LLMAPIHandlerFactory:) -> dict[str, Any]:active_parameters = base_parameters or {}if parameters is None:- parameters = LLMAPIHandlerFactory.get_api_parameters()+ parameters = LLMAPIHandlerFactory.get_api_parameters(llm_config)active_parameters.update(parameters)if llm_config.litellm_params: # type: ignore@@ -261,9 +261,9 @@ class LLMAPIHandlerFactory:return llm_api_handler@staticmethod- def get_api_parameters() -> dict[str, Any]:+ def get_api_parameters(llm_config: LLMConfig | LLMRouterConfig) -> dict[str, Any]:return {- "max_tokens": SettingsManager.get_settings().LLM_CONFIG_MAX_TOKENS,+ "max_tokens": llm_config.max_output_tokens,"temperature": SettingsManager.get_settings().LLM_CONFIG_TEMPERATURE,}diff --git a/skyvern/forge/sdk/api/llm/config_registry.py b/skyvern/forge/sdk/api/llm/config_registry.pyindex 1507ebabf..1f4dc6af4 100644--- a/skyvern/forge/sdk/api/llm/config_registry.py+++ b/skyvern/forge/sdk/api/llm/config_registry.py@@ -76,7 +76,10 @@ if SettingsManager.get_settings().ENABLE_OPENAI:),)LLMConfigRegistry.register_config(- "OPENAI_GPT4O", LLMConfig("gpt-4o", ["OPENAI_API_KEY"], supports_vision=True, add_assistant_prefix=False)+ "OPENAI_GPT4O",+ LLMConfig(+ "gpt-4o", ["OPENAI_API_KEY"], supports_vision=True, add_assistant_prefix=False, max_output_tokens=16384+ ),)LLMConfigRegistry.register_config("OPENAI_GPT4O_MINI",@@ -85,11 +88,18 @@ if SettingsManager.get_settings().ENABLE_OPENAI:["OPENAI_API_KEY"],supports_vision=True,add_assistant_prefix=False,+ max_output_tokens=16384,),)LLMConfigRegistry.register_config("OPENAI_GPT-4O-2024-08-06",- LLMConfig("gpt-4o-2024-08-06", ["OPENAI_API_KEY"], supports_vision=True, add_assistant_prefix=False),+ LLMConfig(+ "gpt-4o-2024-08-06",+ ["OPENAI_API_KEY"],+ supports_vision=True,+ add_assistant_prefix=False,+ max_output_tokens=16384,+ ),)@@ -137,6 +147,7 @@ if SettingsManager.get_settings().ENABLE_ANTHROPIC:["ANTHROPIC_API_KEY"],supports_vision=True,add_assistant_prefix=True,+ max_output_tokens=8192,),)diff --git a/skyvern/forge/sdk/api/llm/models.py b/skyvern/forge/sdk/api/llm/models.pyindex 015ad5745..20b99c17c 100644--- a/skyvern/forge/sdk/api/llm/models.py+++ b/skyvern/forge/sdk/api/llm/models.py@@ -34,6 +34,7 @@ class LLMConfigBase:@dataclass(frozen=True)class LLMConfig(LLMConfigBase):litellm_params: Optional[LiteLLMParams] = field(default=None)+ max_output_tokens: int = SettingsManager.get_settings().LLM_CONFIG_MAX_TOKENS@dataclass(frozen=True)@@ -69,6 +70,7 @@ class LLMRouterConfig(LLMConfigBase):allowed_fails: int | None = Noneallowed_fails_policy: AllowedFailsPolicy | None = Nonecooldown_time: float | None = None+ max_output_tokens: int = SettingsManager.get_settings().LLM_CONFIG_MAX_TOKENSclass LLMAPIHandler(Protocol):