Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Retrosynthesis Prediction with Conditional Graph Logic Network
Authors: Hanjun Dai, Chengtao Li, Connor Coley, Bo Dai, Le Song
NeurIPS 2019 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 6 Experiment Dataset: We mainly evaluate our method on a benchmark dataset named USPTO-50k, which contains 50k reactions of 10 different types in the US patent literature. We use exactly the same training/validation/test splits as Coley et al. [8], which contain 80%/10%/10% of the total 50k reactions. Table 1 contains the detailed information about the benchmark. [...] We present the top-k exact match accuracy in Table 3, where k ranges from {1, 3, 5, 10, 20, 50}. |
| Researcher Affiliation | Collaboration | Hanjun Dai , Chengtao Li2, Connor W. Coley , Bo Dai , Le Song Google Research, Brain Team, EMAIL 2Galixir Inc., EMAIL Massachusetts Institute of Technology, EMAIL Georgia Institute of Technology, Ant Financial, EMAIL |
| Pseudocode | Yes | Algorithm 1 Importance Sampling for br ( ) |
| Open Source Code | Yes | Our code is released at https://github.com/Hanjun Dai/GLN. |
| Open Datasets | Yes | We mainly evaluate our method on a benchmark dataset named USPTO-50k, which contains 50k reactions of 10 different types in the US patent literature. We use exactly the same training/validation/test splits as Coley et al. [8], which contain 80%/10%/10% of the total 50k reactions. |
| Dataset Splits | Yes | We use exactly the same training/validation/test splits as Coley et al. [8], which contain 80%/10%/10% of the total 50k reactions. |
| Hardware Specification | Yes | It takes about 12 hours to train with a single GTX 1080Ti GPU. |
| Software Dependencies | No | The paper mentions software like 'RDKit' and 'rdchiral' ('We use rdchiral [31] to extract the retrosynthesis templates...'), but does not specify their version numbers or the versions of other dependencies like PyTorch, TensorFlow, or Python. |
| Experiment Setup | Yes | We train our model for up to 150k updates with batch size of 64. [...] We tune embedding sizes in {128, 256}, GNN layers {3, 4, 5} and GNN aggregation in {max, mean, sum} using validation set. |