AnirbanSaha/mistral-nemo-12b-tlink-full-autoregressive
AnirbanSaha/mistral-nemo-12b-tlink-full-autoregressive is a 12 billion parameter Mistral-Nemo-Instruct-2407 based model, fine-tuned for temporal relation classification. This model specializes in identifying and categorizing temporal links (BEFORE, AFTER, OTHER, NONE) between events. It achieves an accuracy of 0.7917 and a Macro-F1 of 0.7919 on the fahmidiqbal/tlink-classification dataset, making it suitable for tasks requiring precise temporal ordering.
Loading preview...
Model Overview
AnirbanSaha/mistral-nemo-12b-tlink-full-autoregressive is a 12 billion parameter language model built upon the mistralai/Mistral-Nemo-Instruct-2407 base model. It has undergone full-parameter fine-tuning specifically for temporal relation classification, a task focused on determining the chronological relationship between events.
Key Capabilities
- Temporal Relation Classification: The model is designed to classify temporal links into categories such as BEFORE, AFTER, OTHER, and NONE.
- Specialized Fine-tuning: It leverages the
fahmidiqbal/tlink-classificationdataset for its specialized training, ensuring proficiency in identifying temporal dependencies.
Performance Highlights
Evaluated on its target task, the model demonstrates solid performance:
- Accuracy: 0.7917
- Macro-F1: 0.7919
- Weighted-F1: 0.7929
Notably, it produced 0 invalid outputs out of 192 test cases, indicating reliable classification. Detailed evaluation metrics, including a classification report and confusion matrix, are available in the evaluation/ directory.
Good For
This model is particularly well-suited for applications requiring automated analysis of temporal relationships within text, such as:
- Event ordering in narratives or historical documents.
- Information extraction systems focused on timelines.
- Natural language understanding tasks where precise temporal context is crucial.