My Weird Prompts
Daniel Rosehill
Podcast
Episodes
Listen, download, subscribe
Training a Model When You Don't Know the Labels Yet
What happens when you need to train an extraction model but the label set doesn't exist yet? This episode works through a listener's parametric-search project — scraping vendor spec sheets to build a filterable home inventory of adapters, cables, and power supplies — where the schema grows during annotation itself. We dig into attribute value extraction, why standard NER assumes a closed world, and how schema-conditioned models like GLiNER let you pass labels in as text at inference time with no retraining. Along the way: a 300M-parameter encoder that beats a 13B model on out-of-domain benchmarks, why annotation tooling never caught up to the research, and the anchoring risk hiding inside model-assisted labeling. Key sources: • GLiNER2 README — https://raw.githubusercontent.com/fastino-ai/gliner2/main/README.md • GLiNER2 JSON extraction tutorial — https://raw.githubusercontent.com/fastino-ai/gliner2/main/tutorial/3-json_extraction.md • GLiNER2 training tutorial — https://raw.githubusercontent.com/fastino-ai/gliner2/main/tutorial/9-training.md Episode 5936 — myweirdprompts.com/n/5936 (permanent short link: myweirdprompts.com/120112)
My Weird Prompts RSS Feed
