Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
October 14, 2025
·
Houston
BERT Fine-tuning on MultiNLI
Demonstration of fine‑tuning BERT on MultiNLI for three‑way NLI, covering the training pipeline, class hierarchy, unit tests, and live inference.
Overview
BERT fine-tuning for Natural Language Inference: classifying statement pairs as entailment, contradiction, or neutral. Using a model trained on MultiNLI, I’ll present the training architecture diagram (pipeline components and Python class hierarchy), run live unit tests, and demonstrate real-time inference on example statements.
Links
Tech stack