Knowledgator Releases GLiFormer: A 575M-Parameter Encoder That Hits 91.10 F1 on Nested JSON Extraction Without Generating Tokens

Knowledgator has released GLiFormer, a 575M-parameter encoder designed to extract nested JSON data by grounding values directly in source text rather than generating new tokens. The model achieves a 91.10 F1 score on extraction tasks, providing an efficient alternative for structured data processing that nears the performance of significantly larger language models.