Hierarchical Decision Making by Generating and Following Natural Language Instructions

Authors: Hengyuan Hu, Denis Yarats, Qucheng Gong, Yuandong Tian, Mike Lewis

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We gather a dataset of 76 thousand pairs of instructions and executions from human play, and train instructor and executor models. Experiments show that models generate intermediate plans in natural langauge significantly outperform models that directly imitate human actions.
Researcher Affiliation Collaboration Hengyuan Hu Facebook AI Research hengyuan@fb.com Denis Yarats New York University & Facebook AI Research denisyarats@cs.nyu.edu Qucheng Gong Facebook AI Research qucheng@fb.com Yuandong Tian Facebook AI Research yuandong@fb.com Mike Lewis Facebook AI Research mikelewis@fb.com
Pseudocode No The paper describes the models and methods in text and through diagrams, but does not include any structured pseudocode or algorithm blocks.
Open Source Code Yes Our code is open-sourced at www.github.com/facebookresearch/minirts
Open Datasets Yes We gather a dataset of 76 thousand pairs of instructions and executions from human play, and train instructor and executor models. Our dataset is available. For more details, refer to the Appendix. Our code, models and data.
Dataset Splits No The paper mentions the total dataset size and a frame sampling strategy for training, but does not provide specific percentages, counts, or references to predefined splits for training, validation, and test sets.
Hardware Specification No The paper does not provide specific hardware details such as GPU/CPU models, memory amounts, or types of computing resources used for the experiments.
Software Dependencies No The paper mentions using 'Parl AI [12]' but does not provide specific version numbers for ParlAI or any other software libraries or dependencies used in the experiments.
Experiment Setup Yes For each unit, we consider a history of recent N instructions (N = 5 in all our experiments)... We use the RNN DISCRIMINATIVE instructor with 500 instructions. (Table 2 caption). Instructor Model Negative Log Likelihood Win/Lose/Draw rate (%) (with N instructions) N=50 N=250 N=500 (Table 3 header)