juwon1105/RLCR-qwen3-1.7B-hotpot5000

TEXT GENERATIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:2BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 20, 2026Architecture:Transformer Featherless Exclusive Cold

The juwon1105/RLCR-qwen3-1.7B-hotpot5000 model is a 1.7 billion parameter Qwen3-based causal language model fine-tuned on the HotpotQA dataset. Developed by juwon1105, it leverages the GRPO training method, known for enhancing mathematical reasoning, to specialize in question answering tasks. This model is optimized for generating coherent and contextually relevant responses to complex queries, making it suitable for applications requiring detailed information retrieval and synthesis.

Loading preview...

Model Overview

This model, juwon1105/RLCR-qwen3-1.7B-hotpot5000, is a fine-tuned version of the Qwen3-1.7B base model. It has been specifically trained on the mehuldamani/hotpot_qa dataset, which focuses on multi-hop question answering, requiring reasoning over multiple documents to find an answer.

Key Capabilities

  • Enhanced Question Answering: Specialized in answering complex questions, particularly those requiring information synthesis from multiple sources, due to its fine-tuning on the HotpotQA dataset.
  • GRPO Training Method: Utilizes the GRPO (Gradient-based Reward Optimization) method, as introduced in the DeepSeekMath paper, which is designed to improve reasoning capabilities, particularly in mathematical contexts, but applicable to general reasoning tasks.
  • Qwen3 Architecture: Built upon the Qwen3-1.7B architecture, providing a solid foundation for language understanding and generation.

Training Details

The model was trained using the TRL (Transformer Reinforcement Learning) library. The application of GRPO during training aims to optimize the model's ability to generate accurate and well-reasoned answers. This approach distinguishes it from models trained with standard supervised fine-tuning by focusing on reward-based optimization.

Good For

  • Applications requiring detailed and reasoned answers to complex questions.
  • Information retrieval systems where synthesizing information from various points is crucial.
  • Research into the effectiveness of GRPO for general question-answering tasks.

Finetunes
1 models