Fixed context usage calculation
This commit is contained in:
@@ -25,14 +25,6 @@ mimetypes.add_type("application/javascript", ".jsx")
|
||||
mimetypes.add_type("text/javascript", ".js")
|
||||
mimetypes.add_type("text/javascript", ".jsx")
|
||||
|
||||
class TestLLM:
|
||||
def infer(self, prompt: str, context: str):
|
||||
yield "<reasoning>"
|
||||
time.sleep(2)
|
||||
yield "test reasoning"
|
||||
time.sleep(2)
|
||||
yield "</reasoning>"
|
||||
|
||||
class Main:
|
||||
@classmethod
|
||||
async def create(cls, config: Config):
|
||||
@@ -44,20 +36,24 @@ class Main:
|
||||
|
||||
match self._config.llm_engine:
|
||||
case "local":
|
||||
self._llm = LocalLlmEngine(self._config.model)
|
||||
self._llm = LocalLlmEngine(
|
||||
self._config.model,
|
||||
self._config.temperature,
|
||||
self._config.token_limit
|
||||
)
|
||||
case "hf":
|
||||
self._llm = HfLlmEngine(
|
||||
model_id=self._config.model,
|
||||
api_token=self._config.api_token,
|
||||
temperature=self._config.temperature
|
||||
self._config.model,
|
||||
self._config.temperature,
|
||||
self._config.api_token,
|
||||
)
|
||||
case "openai":
|
||||
self._llm = OpenAILlmEngine(
|
||||
model=self._config.model,
|
||||
api_key=self._config.api_token
|
||||
self._config.model,
|
||||
self._config.temperature,
|
||||
self._config.api_token,
|
||||
self._config.token_limit
|
||||
)
|
||||
case "test":
|
||||
self._llm = TestLLM()
|
||||
case _:
|
||||
raise ValueError(f"Invalid LLM engine: {self._config.llm_engine}")
|
||||
self._io_buffer = WebIOBuffer()
|
||||
@@ -65,7 +61,7 @@ class Main:
|
||||
system_prompt=self._system_prompt,
|
||||
action_schema=self._action_schema,
|
||||
working_memory=WorkingMemory(),
|
||||
system_metrics=SystemMetrics(),
|
||||
metrics=SystemMetrics(),
|
||||
llm=self._llm,
|
||||
validator=XMLValidator(self._action_schema),
|
||||
parser=ResponseParser(self._io_buffer)
|
||||
|
||||
Reference in New Issue
Block a user