Structured agents for physical construction
Authors: Victor Bapst, Alvaro Sanchez-Gonzalez, Carl Doersch, Kimberly Stachenfeld, Pushmeet Kohli, Peter Battaglia, Jessica Hamrick
ICML 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We ran experiments to evaluate the effectiveness of different agent architectures (see Table 1) on our construction tasks |
| Researcher Affiliation | Industry | 1Deep Mind, London, UK. Correspondence to: Jessica Hamrick <jhamrick@google.com>. |
| Pseudocode | No | The paper does not contain any clearly labeled 'Pseudocode' or 'Algorithm' blocks. |
| Open Source Code | No | The paper provides a URL for videos of agent behaviors (https://tinyurl.com/y7wtfen9) but does not contain an explicit statement or link indicating the public release of the source code for its methodology. |
| Open Datasets | No | The experiments use a 'procedurally-generated 2D world implemented in Unity with the Box2D physics engine', implying the data is generated on the fly rather than being a pre-existing, publicly available dataset with a specific access link or formal citation. |
| Dataset Splits | No | The paper describes training with a curriculum and evaluation on 10,000 scenes, but it does not explicitly mention or quantify a validation dataset split or a specific validation methodology. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., CPU, GPU models, or cloud computing instances) used for running the experiments. |
| Software Dependencies | No | The paper mentions 'Unity (Juliani et al., 2018)' and the 'Box2D physics engine (Catto, 2013)' as components of the simulated environment, but it does not specify version numbers for these or any other software dependencies crucial for replication. |
| Experiment Setup | Yes | For efficient training, we found it was important to apply a curriculum which progressively increases the complexity of the task across training episodes. |