VGLee/qwen1.5-4b-universal_ner-galore
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Apr 26, 2024License:otherArchitecture:Transformer Warm
VGLee/qwen1.5-4b-universal_ner-galore is a 4 billion parameter language model, fine-tuned from Qwen/Qwen1.5-4B, specifically optimized for Universal Named Entity Recognition (NER) tasks. This model leverages a 32768 token context length and is designed to excel in identifying and classifying entities across various text types. It is particularly suited for applications requiring robust and generalized NER capabilities.
Loading preview...
Model Overview
VGLee/qwen1.5-4b-universal_ner-galore is a specialized 4 billion parameter model derived from the Qwen1.5-4B architecture. Its primary focus is Universal Named Entity Recognition (NER), having been fine-tuned on the universal_ner_all dataset.
Key Capabilities
- Universal Named Entity Recognition: Optimized for identifying and classifying diverse entities within text, making it suitable for a broad range of NER applications.
- Qwen1.5 Base: Built upon the robust Qwen1.5-4B foundation, inheriting its general language understanding capabilities.
- Training Configuration: Trained with a learning rate of 1e-05, a batch size of 2 (accumulated to 16), and a cosine learning rate scheduler over 1 epoch, indicating a focused fine-tuning approach for NER.
Good For
- Information Extraction: Ideal for tasks requiring the automated extraction of specific entities (e.g., persons, organizations, locations, dates) from unstructured text.
- Data Annotation & Pre-processing: Can be used to pre-annotate datasets or enhance data quality by automatically identifying key entities.
- Research in NER: Provides a fine-tuned base for further experimentation and development in universal NER systems.