This description was created by Ellipsis for commit 50ea313466c75883de0f63ba351df75e122ee5dc
Summary:
This PR introduces cooldown and failure handling configurations to the LLM API handler in skyvern/forge/sdk/api/llm by updating LLMRouterConfig and get_llm_api_handler_with_router.
Key points:
Added disable_cooldowns, allowed_fails, allowed_fails_policy, and cooldown_time to LLMRouterConfig in skyvern/forge/sdk/api/llm/models.py.
Updated get_llm_api_handler_with_router in skyvern/forge/sdk/api/llm/api_handler_factory.py to use new cooldown configurations.
Integrated AllowedFailsPolicy from litellm for handling failure policies.
Summary:
This PR introduces cooldown and failure handling configurations to the LLM API handler in
skyvern/forge/sdk/api/llmby updatingLLMRouterConfigandget_llm_api_handler_with_router.Key points:
disable_cooldowns,allowed_fails,allowed_fails_policy, andcooldown_timetoLLMRouterConfiginskyvern/forge/sdk/api/llm/models.py.get_llm_api_handler_with_routerinskyvern/forge/sdk/api/llm/api_handler_factory.pyto use new cooldown configurations.AllowedFailsPolicyfromlitellmfor handling failure policies.Generated with :heart: by ellipsis.dev