Neural Programmer-Interpreters

Authors: Scott Reed, Nando de Freitas

ICLR 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the capability of our model to learn several types of compositional programs: addition, sorting, and canonicalizing 3D models.
Researcher Affiliation Industry Scott Reed & Nando de Freitas Google Deep Mind London, UK scott.ellison.reed@gmail.com nandodefreitas@google.com
Pseudocode Yes Algorithm 1 Neural programming inference
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes For training data, we used renderings of the 3D car CAD models from (Fidler et al., 2012).
Dataset Splits No The paper describes training data sizes and testing on unseen data, but does not provide specific details on validation dataset splits by percentages, counts, or references to predefined validation sets.
Hardware Specification No The paper mentions aspects like 'device memory' and 'LSTM layers' but does not provide specific hardware details such as GPU/CPU models or memory amounts used for experiments.
Software Dependencies No The paper mentions the 'ADAM solver (Kingma & Ba, 2015)' but does not provide specific version numbers for software libraries or frameworks used.
Experiment Setup Yes We trained the NPI using the ADAM solver (Kingma & Ba, 2015) with base learning rate 0.0001, batch size 1, and decayed the learning rate by a factor of 0.95 every 10,000 steps.