patilshrinivas/Qwen2.5-3B-Instruct-drug-ade-relation-extractor

TEXT GENERATIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:3.1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 6, 2026Architecture:Transformer Featherless Exclusive Cold

The patilshrinivas/Qwen2.5-3B-Instruct-drug-ade-relation-extractor is a 3.1 billion parameter Qwen2.5-3B-Instruct model fine-tuned by patilshrinivas. It specializes in extracting drug-adverse drug event (ADE) relations from clinical sentences, leveraging QLoRA fine-tuning on the ADE Corpus V2. This model is optimized for structured clinical information extraction, achieving a Strict Pair F1 of 71.98% and a JSON Validity of 99.00%.

Loading preview...

Overview

This model, patilshrinivas/Qwen2.5-3B-Instruct-drug-ade-relation-extractor, is a specialized version of the Qwen/Qwen2.5-3B-Instruct base model, fine-tuned using QLoRA for the specific task of extracting drug-adverse drug event (ADE) relations from clinical text. It was trained on the ADE Corpus V2 dataset, utilizing 4-bit NF4 quantization and BF16 compute dtype during fine-tuning.

Key Capabilities

  • ADE Relation Extraction: Specifically designed to identify and extract relationships between drugs and adverse drug events from clinical sentences.
  • High JSON Validity: Achieves 99.00% JSON validity, ensuring structured and parseable output for downstream applications.
  • Strong Performance: Demonstrates robust performance with a Strict Pair F1 score of 71.98%, a Drug F1 of 92.82%, and an Effect F1 of 76.44%.
  • G-Eval Score: Reports an average correctness score of 0.8749 from G-Eval using gpt-4o-mini.

Intended Use

This model is primarily intended for research and experimentation in structured clinical information extraction. It is crucial to note that it is not a medical diagnostic system and should not be used as a substitute for professional clinical judgment.