Yasaman Bahri
Research Scientist · Harvard University
Yasaman Bahri is a research scientist working on neural networks, stochastic gradient optimization and topological materials.
About
Yasaman Bahri is a research scientist working on neural networks, stochastic gradient optimization and topological materials. Her Harvard talk derives geometry in language models from word2vec co-occurrence statistics, using a hypercube model for analogies and a lattice model for winding manifolds.
Same subject, AI & Machine Learning
Connections
4Their places, courses, books and lectures.
