TAUR-dev/rankalign-v6-gemma-2-2b-d0.15-e2-hc-b2d-dbl-all-tco-ln-nv1-ng1-fsx

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-tco-ln-nv1-ng1-fsx is a 2.6 billion parameter language model fine-tuned from Google's Gemma-2-2b base model. This model is specifically optimized for hypernym prediction tasks, trained using the rankalign project's methodology with online typicality correction and length normalization. It is designed to excel in identifying hierarchical relationships between concepts, particularly within the 'hypernym-concat-bananas-to-dogs-double-all' task. Its specialized fine-tuning makes it suitable for research and applications requiring precise hypernym generation and validation.

Loading preview...

Model Overview

This model, rankalign-v6-gemma-2-2b-d0.15-e2-hc-b2d-dbl-all-tco-ln-nv1-ng1-fsx, is a fine-tuned checkpoint derived from the google/gemma-2-2b base model, developed as part of the rankalign project. It features 2.6 billion parameters and has been trained for 2 epochs with a delta of 0.15.

Key Training Details

The model's training focused on a specific task identified as hypernym-concat-bananas-to-dogs-double-all. Notable training parameters include:

  • Base Model: google/gemma-2-2b
  • Version: v6
  • Epochs: 2
  • Delta: 0.15
  • Typicality Correction: Online
  • Length Normalization: Enabled
  • Preference Loss Weight: 1
  • NLL Validator Weight: 1
  • NLL Generator Weight: 1

Intended Use Cases

This model is specifically designed for tasks involving hypernym prediction and validation. Its fine-tuning process, which includes online typicality correction and length normalization, suggests an emphasis on generating semantically accurate and contextually appropriate hypernyms. It is particularly well-suited for research and development in semantic hierarchy understanding and knowledge graph construction, especially for the specific hypernym tasks it was trained on, such as those involving 'bananas' to 'dogs' and other listed categories.