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

MarkTechPost
Read full post
Knowledgator Engineering launched GLiFormer, a 575M-parameter encoder that performs multiple information extraction tasks including nested JSON structuring without token-by-token output generation. It supports named-entity recognition, classification, relation extraction, and embeddings with schema-conditioned inference and is available under Apache 2.0 license.

More in Machine Learning

Machine Learning3 min read

Arcee AI built Trinity for about $20m, and investors now value it at $1bn+

The Next Web
Machine Learning4 min read

Workiva puts business outcomes at the center of AI-native workflows

SiliconANGLE
Machine Learning4 min read

Perplexity’s AI agents helped build a database. They weren’t allowed to run it.

The New Stack (AI)