vodorod/Dorod-Sky
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

shu/remove max completion tokens for ollama (#2300)

Browse Source
This commit is contained in:
Shuchang Zheng
2025-05-06 12:21:42 -07:00
committed by GitHub
parent 5971108237
commit 349e506dc1
2 changed files with 6 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
skyvern/forge/sdk/api/llm/api_handler_factory.py
View File

@@ -430,10 +430,12 @@ class LLMAPIHandlerFactory:
@staticmethod
def get_api_parameters(llm_config: LLMConfig | LLMRouterConfig) -> dict[str, Any]:
params: dict[str, Any] = {}
if llm_config.max_completion_tokens is not None:
params["max_completion_tokens"] = llm_config.max_completion_tokens
elif llm_config.max_tokens is not None:
params["max_tokens"] = llm_config.max_tokens
if not llm_config.model_name.startswith("ollama/"):
# OLLAMA does not support max_completion_tokens
if llm_config.max_completion_tokens is not None:
params["max_completion_tokens"] = llm_config.max_completion_tokens
elif llm_config.max_tokens is not None:
params["max_tokens"] = llm_config.max_tokens
if llm_config.temperature is not None:
params["temperature"] = llm_config.temperature

1
skyvern/forge/sdk/api/llm/config_registry.py
View File

@@ -815,7 +815,6 @@ if settings.ENABLE_OLLAMA:
["OLLAMA_SERVER_URL", "OLLAMA_MODEL"],
supports_vision=False, # Ollama does not support vision yet
add_assistant_prefix=False,
max_completion_tokens=settings.LLM_CONFIG_MAX_TOKENS,
litellm_params=LiteLLMParams(
api_base=settings.OLLAMA_SERVER_URL,
api_key=None,