TAUR-dev/rankalign-v6-gemma-2-2b-d0.15-e2-hc-b2d-dbl-all-tcs-fsx-lo0.1

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:2.6BQuant:BF16Ctx Length:8kPublished:Apr 6, 2026Architecture:Transformer Warm

TAUR-dev/rankalign-v6-gemma-2-2b-d0.15-e2-hc-b2d-dbl-all-tcs-fsx-lo0.1 is a 2.6 billion parameter language model fine-tuned from Google's Gemma-2-2b base model. Developed by TAUR-dev as part of the rankalign project, this model is specifically optimized for hypernym-concat-bananas-to-dogs-double-all tasks. It incorporates a preference loss weight of 1 and self-typicality correction, making it suitable for specialized linguistic relation extraction and classification.

Loading preview...

Model Overview

This model, rankalign-v6-gemma-2-2b-d0.15-e2-hc-b2d-dbl-all-tcs-fsx-lo0.1, is a fine-tuned checkpoint derived from the google/gemma-2-2b base model. It is part of the rankalign project, focusing on specific linguistic tasks.

Training Details

The model underwent two training epochs with a delta of 0.15. Key training parameters include:

  • Task: hypernym-concat-bananas-to-dogs-double-all
  • Preference Loss Weight: 1
  • NLL Validator/Generator Weight: 0
  • Typicality Correction: Self
  • Force Same-X: True
  • Labeled-only Ratio: 0.1

Use Cases

This model is particularly suited for research and applications involving hypernym relation extraction and classification, as indicated by its training task. The provided evaluation scripts demonstrate its application across various hypernym-related datasets, such as hypernym-bananas, hypernym-dogs, and hypernym-chairs.