MCPcopy Create free account
hub / github.com/InternLM/InternBootcamp / AgentLoopOutput

Class AgentLoopOutput

verl/verl/experimental/agent_loop/agent_loop.py:135–157  ·  view source on GitHub ↗

Agent loop output.

Source from the content-addressed store, hash-verified

133
134
135class AgentLoopOutput(BaseModel):
136 """Agent loop output."""
137
138 prompt_ids: list[int]
139 """Prompt token ids."""
140 response_ids: list[int]
141 """Response token ids including LLM generated token, tool response token."""
142 response_mask: list[int]
143 """Response mask, 1 for LLM generated token, 0 for tool response token."""
144 response_logprobs: Optional[list[float]] = None
145 """Log probabilities for the response tokens."""
146 routed_experts: Optional[Any] = None
147 """Routed experts for the total tokens."""
148 multi_modal_data: Optional[dict[str, Any]] = None
149 """Multi-modal data for multi-modal tools."""
150 reward_score: Optional[float] = None
151 """Reward score for the trajectory."""
152 num_turns: int = 0
153 """Number of chat turns, including user, assistant, tool."""
154 metrics: AgentLoopMetrics
155 """Auxiliary performance metrics"""
156 extra_fields: dict[str, Any] = {}
157 """Extra fields for dynamic addition."""
158
159
160class _InternalAgentLoopOutput(AgentLoopOutput):

Callers 6

runMethod · 0.90
runMethod · 0.90
runMethod · 0.90
runMethod · 0.90

Calls

no outgoing calls

Tested by

no test coverage detected