AnirbanSaha/qwen2.5-14b-tlink-sentence-only-autoregressive
The AnirbanSaha/qwen2.5-14b-tlink-sentence-only-autoregressive is a 14.8 billion parameter language model, based on Qwen2.5-14B-Instruct, specifically fine-tuned for temporal relation classification. It excels at identifying 'BEFORE', 'AFTER', 'OTHER', and 'NONE' relationships between events within sentences. This model is optimized for tasks requiring precise temporal understanding, achieving an accuracy of 80.21% and a Macro-F1 of 80.06% on its specialized dataset.
Loading preview...
Overview
This model, AnirbanSaha/qwen2.5-14b-tlink-sentence-only-autoregressive, is a specialized variant of the Qwen2.5-14B-Instruct base model, featuring 14.8 billion parameters and a 32768-token context length. It has undergone full-parameter fine-tuning specifically for temporal relation classification within sentences.
Key Capabilities
- Temporal Relation Classification: Identifies the temporal relationship between events or entities in text.
- Specific Labeling: Classifies relationships into predefined categories:
BEFORE,AFTER,OTHER, andNONE. - High Accuracy: Achieves an accuracy of 0.8021 and a Macro-F1 score of 0.8006 on its dedicated evaluation dataset.
- Robust Output: Demonstrated 0 invalid outputs during testing, indicating reliable classification.
Good for
- Event Ordering: Applications requiring precise ordering of events mentioned in text.
- Information Extraction: Extracting temporal dependencies from unstructured text data.
- Natural Language Understanding (NLU): Enhancing NLU systems with fine-grained temporal awareness.
- Specialized NLP Tasks: Use cases where identifying 'BEFORE' or 'AFTER' relationships is critical, such as timeline generation or historical document analysis.