Enhancing Robot Program Synthesis Through Environmental Context

Authors: Tianyi Chen, Qidi Wang, Zhen Dong, Liwei Shen, Xin Peng

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experimental evaluations and ablation studies on the partially observed Viz Doom domain authenticate that our method offers superior generalization capability across various tasks and greater robustness when encountering noises.
Researcher Affiliation Academia Tianyi Chen Fudan University tianychen21@m.fudan.edu.cn Qidi Wang Fudan University 21210240038@m.fudan.edu.cn Zhen Dong Fudan University zhendong@fudan.edu.cn Liwei Shen Fudan University shenliwei@fudan.edu.cn Xin Peng Fudan University pengxin@fudan.edu.cn
Pseudocode No The paper includes architectural diagrams and a Domain-Specific Language (DSL) description (Figure 1 and Figure 2), but no structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any concrete information or links for open-source code availability.
Open Datasets No To generate a dataset for learning environmental-context embeddings, we adopt the same approach as previous studies [14, 47], randomly generating 100, 000 distinct samples. The paper describes generating its own dataset based on an existing approach, but does not provide any link, DOI, or explicit statement that this specific generated dataset is publicly available.
Dataset Splits Yes The dataset is then partitioned into a training set with 70, 000 samples, a validation set with 20, 000 samples, and a testing set with 10, 000 samples.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions various neural network architectures (e.g., convolutional network layers, RNN, Transformer) and frameworks like Viz Doom, but does not provide specific version numbers for any software dependencies.
Experiment Setup No The paper states that 'The supplementary material contains the detailed parameters of all the compared methods,' but the main text does not provide specific experimental setup details such as hyperparameter values, batch sizes, or training configurations.