Deep Learning with Dynamic Computation Graphs
Authors: Moshe Looks, Marcello Herreshoff, DeLesley Hutchins, Peter Norvig
ICLR 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We present empirical results for our implementation in Tensor Flow. Section 3 presents a combinator library for concisely implementing models with DCGs using dynamic batching. The test results shown in Table 1 emphasize the importance of batching, especially on GPUs. |
| Researcher Affiliation | Industry | Moshe Looks, Marcello Herreshoff, De Lesley Hutchins & Peter Norvig Google Inc. {madscience,marcelloh,delesley,pnorvig}@google.com |
| Pseudocode | No | The paper describes algorithms conceptually but does not provide pseudocode blocks or sections explicitly labeled as "Algorithm". |
| Open Source Code | Yes | 1The library is called Tensor Flow Fold and lives at http://github.com/tensorflow/fold. |
| Open Datasets | Yes | We used constituency Tree-LSTMs with tuned Glove vectors for word embedding... Results are shown in Table 2, including the best previously reported results. Fine-grained accuracy is measured for all trees and calculated based on the five possible labels. Test set accuracies on the Stanford Sentiment Treebank |
| Dataset Splits | Yes | Noting the small size of this dataset (8544/1101/2210 trees for train/dev/test) |
| Hardware Specification | Yes | The CPU tests were run on a Dell z620 workstation with dual 8-core Intel Xeon processors (32 hardware threads), and the GPU tests were done using a consumer Nvidia Ge Force GTX-1080 card. |
| Software Dependencies | No | The paper mentions using "Tensor Flow" and the library "Tensor Flow Fold" but does not specify exact version numbers for these software dependencies, which would be necessary for full reproducibility. |
| Experiment Setup | Yes | The tree size is 128, with a state size of 1024 for the LSTM. We used constituency Tree-LSTMs with tuned Glove vectors for word embedding... and increased the LSTM state size from 150 to 300, leaving all other hyperparameters unchanged. |