SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.11978 · 2024

Dialogue Action Tokens: Steering Language Models in Goal-Directed Dialogue with a Multi-Turn Planner

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 8 functions out of this paper's own repositories and ran 7 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
likenneth/dialogue_action_token canonical 7 of 8
FunctionStatusWhere it lives
eval_actor Ran likenneth/dialogue_action_token/dat/pre_bc.py
code served (permissive licence) · get_code("ca55c2bc3a455a58")
extract_leading_int Ran likenneth/dialogue_action_token/envs/redteam_env.py
code served (permissive licence) · get_code("ff5a465969609b7b")
format_docstring Ran likenneth/dialogue_action_token/envs/sotopia_utils/utils.py
code served (permissive licence) · get_code("f90433d593217a49")
get_bio Ran likenneth/dialogue_action_token/envs/sotopia_utils/utils.py
code served (permissive licence) · get_code("aae3c57500fd1a36")
load_pickle Ran likenneth/dialogue_action_token/dat/td3.py
code served (permissive licence) · get_code("38d71c4ed76282b3")
preprocess_function Ran likenneth/dialogue_action_token/dat/utils.py
code served (permissive licence) · get_code("73e2c40bff6225ae")
preprocess_no_padding Ran likenneth/dialogue_action_token/dat/utils.py
code served (permissive licence) · get_code("db147f8b7e6fcb02")
load_data Not yet run likenneth/dialogue_action_token/dat/bc.py
code served (permissive licence) · get_code("3e6d732bbfaddc55")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We present an approach called Dialogue Action Tokens (DAT) that adapts language model agents to plan goal-directed dialogues. The core idea is to treat each utterance as an action, thereby converting dialogues into games where existing approaches such as reinforcement learning can be applied. Specifically, we freeze a pretrained language model and train a small planner model that predicts a continuous action vector, used for controlled generation in each round. This design avoids the problem of language degradation under reward optimization. When evaluated on the Sotopia platform for social simulations, the DAT-steered LLaMA model surpasses GPT-4's performance. We also apply DAT to steer an attacker language model in a novel multi-turn red-teaming setting, revealing a potential new attack surface.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2406.11978")
get_code_for_paper("2406.11978")
have("2406.11978")

Connect an agent — have() is free.