|
|
@ -217,9 +217,9 @@ class OpenLLM(LLM):
|
|
|
|
def _identifying_params(self) -> IdentifyingParams:
|
|
|
|
def _identifying_params(self) -> IdentifyingParams:
|
|
|
|
"""Get the identifying parameters."""
|
|
|
|
"""Get the identifying parameters."""
|
|
|
|
if self._client is not None:
|
|
|
|
if self._client is not None:
|
|
|
|
self.llm_kwargs.update(self._client.configuration)
|
|
|
|
self.llm_kwargs.update(self._client._config())
|
|
|
|
model_name = self._client.model_name
|
|
|
|
model_name = self._client._metadata()["model_name"]
|
|
|
|
model_id = self._client.model_id
|
|
|
|
model_id = self._client._metadata()["model_id"]
|
|
|
|
else:
|
|
|
|
else:
|
|
|
|
if self._runner is None:
|
|
|
|
if self._runner is None:
|
|
|
|
raise ValueError("Runner must be initialized.")
|
|
|
|
raise ValueError("Runner must be initialized.")
|
|
|
@ -265,7 +265,9 @@ class OpenLLM(LLM):
|
|
|
|
self._identifying_params["model_name"], **copied
|
|
|
|
self._identifying_params["model_name"], **copied
|
|
|
|
)
|
|
|
|
)
|
|
|
|
if self._client:
|
|
|
|
if self._client:
|
|
|
|
res = self._client.query(prompt, **config.model_dump(flatten=True))
|
|
|
|
res = self._client.generate(
|
|
|
|
|
|
|
|
prompt, **config.model_dump(flatten=True)
|
|
|
|
|
|
|
|
).responses[0]
|
|
|
|
else:
|
|
|
|
else:
|
|
|
|
assert self._runner is not None
|
|
|
|
assert self._runner is not None
|
|
|
|
res = self._runner(prompt, **config.model_dump(flatten=True))
|
|
|
|
res = self._runner(prompt, **config.model_dump(flatten=True))
|
|
|
@ -300,9 +302,10 @@ class OpenLLM(LLM):
|
|
|
|
self._identifying_params["model_name"], **copied
|
|
|
|
self._identifying_params["model_name"], **copied
|
|
|
|
)
|
|
|
|
)
|
|
|
|
if self._client:
|
|
|
|
if self._client:
|
|
|
|
res = await self._client.acall(
|
|
|
|
async_client = openllm.client.AsyncHTTPClient(self.server_url)
|
|
|
|
"generate", prompt, **config.model_dump(flatten=True)
|
|
|
|
res = (
|
|
|
|
)
|
|
|
|
await async_client.generate(prompt, **config.model_dump(flatten=True))
|
|
|
|
|
|
|
|
).responses[0]
|
|
|
|
else:
|
|
|
|
else:
|
|
|
|
assert self._runner is not None
|
|
|
|
assert self._runner is not None
|
|
|
|
(
|
|
|
|
(
|
|
|
|