viamr-project/qwen3-1.7b-amr-augmented-20260214-1807
The viamr-project/qwen3-1.7b-amr-augmented-20260214-1807 is a 1.7 billion parameter language model developed by viamr-project, fine-tuned for Abstract Meaning Representation (AMR) parsing. This model leverages Reinforcement Learning (veRL) for its training, specializing in converting natural language sentences into their AMR graph representations. With a context length of 40960 tokens, it achieves an F1 score of 82.03 on its benchmark, making it suitable for applications requiring precise semantic parsing.
Loading preview...
Overview
The viamr-project/qwen3-1.7b-amr-augmented-20260214-1807 is a 1.7 billion parameter language model specifically designed and trained for Abstract Meaning Representation (AMR) parsing. Developed by viamr-project, this model utilizes a Reinforcement Learning (veRL) framework to enhance its ability to convert English sentences into structured AMR graphs.
Key Capabilities
- AMR Parsing: Specializes in transforming natural language into Abstract Meaning Representation, providing a deep semantic understanding of sentences.
- Reinforcement Learning (veRL) Trained: Benefits from a reinforcement learning approach, which typically leads to improved performance on specific, complex tasks like semantic parsing.
- Competitive Performance: Achieves an F1 score of 82.03, with Precision at 82.83 and Recall at 81.24 on its internal benchmark, indicating strong performance in AMR conversion.
- Large Context Window: Supports a context length of 40960 tokens, allowing it to process longer inputs for AMR parsing.
Good For
- Semantic Analysis: Ideal for applications requiring detailed semantic understanding of text.
- Natural Language Understanding (NLU): Useful in NLU pipelines where converting text to a structured, machine-readable semantic representation is crucial.
- Knowledge Graph Construction: Can serve as a foundational component for building knowledge graphs from unstructured text by extracting predicate-argument structures.
- Research in AMR: A valuable tool for researchers and developers working on AMR-related tasks and advancements.