Pith. sign in

← back to paper

Review history

arxiv: 2607.10428 · 2 revisions

Enjoy Your Talk: A Human-Centered Benchmark for Multi-Turn Dialogue with Decoupled User Simulation, Target Modeling, and Judging

  1. 2026-08-02 CONDITIONAL MODERATE v1.3.0-alltime-deepseek novelty 6.0
    191950 ms 24313 in 18240 out 2026-08-02T07:14:43.483714+00:00
  2. 2026-07-14 CONDITIONAL HIGH v1.1.0-grok45 novelty 6.0
    57890 ms 30573 in 4013 out 2026-07-14T11:48:25.953017+00:00