multi-turn

This is the official GitHub repository for our survey paper "Beyond Single-Turn: A Survey on Multi-Turn Interactions with Large Language Models".

Generate conversational, tool-calling, structured-output, and preference datasets — easily and at scale

Reproduce alpaca

Pytorch Implementation of "Sequential Dialogue Context Modeling for Spoken Language Understanding( https://arxiv.org/abs/1705.03455 )"

Entity extraction using BERT + CRF for single-tun / multi-turn setting in dialogues

[EMNLP 2025 main] C3 Benchmark: A Bilingual Benchmark for Spoken Dialogue Models Exploring Challenges in Complex Conversations

collect the open dialog corpus and some useful data processing utils.

EMNLP 2019: Dually Interactive Matching Network for Personalized Response Selection in Retrieval-Based Chatbots

MTalk-Bench: Evaluating Speech-to-Speech Models in Multi-Turn Dialogues via Arena-style and Rubrics Protocols

This repo investigates LLMs' tendency to exhibit acquiescence bias in sequential QA interactions. Includes evaluation methods, datasets, benchmarks, and experiment code to assess and mitigate vulnerabilities in conversational consistency and robustness, offering a reproducible framework for future research.

rswarm is a Rust library crafted to streamline AI agent interactions, particularly when working with OpenAI’s API.

[ACL 2026] Official implementation of MTRouter, a cost-aware multi-turn LLM routing framework accepted to ACL 2026 Main Conference.

A smart AI assistant for clinics using LangGraph, LangChain, and OpenAI GPT. It handles patient queries, schedules appointments, detects emergencies, and updates medical records through multi-turn, memory-aware conversations.

[NeurIPS 2025 Spotlight] InterMT: Multi-Turn Interleaved Preference Alignment with Human Feedback

The codes of our paper When to Talk: Chatbot Controls the Timing of Talking during Multi-turn Open-domain Dialogue Generation

AskBench: LLM question-asking/clarification benchmark & dataset with evaluation and training code (paper: arXiv 2602.11199).

A benchmark evolving framework and a benchmark for LLMs' multi-turn instruction following evaluation.

Recovers the original goal of a long Claude Code or Codex session from its first user turn, so you can restate it before continuing — plus a prompt wrapper that adds evidence and hedging constraints to evaluation prompts. Deterministic, local, no model calls. Experimental.

MultiTurnResponseSelection