factor_agent.web.run_view.LLMCall

class LLMCall(number, role, model, round, group, slots, time, seconds, tokens, truncated, empty, prompt, reply, reasoning)

Bases: object

One LLM call of a run, in full, as its call file holds it.

number is the number of its call file. round, group and slots are the round, proposal group and slots it concerned, as the events file says (None and empty without one). time is when it came back (UTC) and seconds how long it took (None without an events file). tokens, truncated (cut off at the output limit, None when the provider’s reply does not say) and empty are read from the provider’s reply. reasoning is the model’s reasoning when the provider returns it, None otherwise.

Attributes

LLMCall.number

LLMCall.role

LLMCall.model

LLMCall.round

LLMCall.group

LLMCall.slots

LLMCall.time

LLMCall.seconds

LLMCall.tokens

LLMCall.truncated

LLMCall.empty

LLMCall.prompt

LLMCall.reply

LLMCall.reasoning