Scenario-Actor：让模型读懂对话场景再自动调采样温度
Read the Room Before Your Model Generates
让模型先读懂场景再自动调采样温度，Llama-3.1-8B 上验证有效，搞解码优化的可以看看。
论文提出 Scenario-Actor 框架，从 Transformer 隐状态中提取场景相关结构，把交际情境编码成几何化的场景嵌入，再据此确定解码温度，替代固定全局温度或 token 级校准。在 Llama-3.1-8B-Instruct 上，Scenario-Actor 在交互式社交场景的目标完成任务中表现强劲，开放式指令遵循有明显增益。与固定温度、token 级校准等多个基线相比，多轮对话成绩也保持有竞争力。
Read the Room Before Your Model Generates
Large language models (LLMs) generally rely on sampling temperature to regulate the amount of randomness in their outputs. Mainstream methods for selecting temperature typically involve fixed, heuristic global settings and context or token-level calibration, rather than a geometrically interpretable representation of the communicative scenario. In this work, we introduce a simple and interpretable temperature selection framework, the \textbf{Scenario-Actor}, which adapts decoding to the inferred situational characteristics. Scenario-Actor extracts scenario-relevant structures from Transformer hidden states and determines the temperature from scenario embedding, a structured geometric representation of the communicative setting. Empirical results show that on Llama-3.1-8B-Instruct, Scenario-Actor achieves strong goal-completion performance in interactive social scenarios, significant gains on open-ended instruction following, and competitive performance on multi-turn conversation compared to multiple baselines.
