PAM Concept Discovery — Pre-computed Data (Model, Centroids, Cluster Assignments, Labels)
Abstract
Pre-computed outputs for "From Topic to Transition Structure: Unsupervised Concept Discovery at Corpus Scale via Predictive Associative Memory." Contains the trained association model checkpoint (29.4M parameters, 150 epochs), k-means centroids at 6 granularities (k=50 to k=2,000), cluster assignments for 24.96M passages, and LLM-generated cluster labels for all 3,900 clusters. These files enable reproduction of evaluation results without retraining. Code at https://github.com/EridosAI/PAM-Concept-Discovery
Fields of Study & Subjects
Topic Modeling
Advanced Clustering Algorithms Research
Computational and Text Analysis Methods
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 15 March 2026
- Language
- EN
- Document type
- Dataset