ruberri/Qwen3-0.6B-mcqa-noreason-phase3

Hugging Face
TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.8BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jun 3, 2025Architecture:Transformer Featherless Exclusive Warm

ruberri/Qwen3-0.6B-mcqa-noreason-phase3 is a 0.8 billion parameter language model, fine-tuned from ruberri/Qwen3-0.6B-mcqa-reason-phase2. This model is specifically trained for multiple-choice question answering (MCQA) without requiring reasoning, leveraging a 32K context length. It is optimized for direct answer retrieval in MCQA tasks, making it suitable for applications needing efficient, non-reasoning-based question answering.

Loading preview...

Model Overview

ruberri/Qwen3-0.6B-mcqa-noreason-phase3 is a 0.8 billion parameter language model, fine-tuned from the ruberri/Qwen3-0.6B-mcqa-reason-phase2 base model. This iteration focuses on multiple-choice question answering (MCQA) tasks where explicit reasoning is not required for the answer. It utilizes a substantial 32,768 token context length, allowing it to process longer inputs for question answering.

Key Capabilities

  • Multiple-Choice Question Answering (MCQA): Specialized in answering multiple-choice questions directly.
  • Non-Reasoning Focus: Optimized for scenarios where answers can be derived without complex logical inference.
  • Large Context Window: Supports a 32K context length, beneficial for understanding detailed questions or passages.

Training Details

The model was fine-tuned using the TRL (Transformer Reinforcement Learning) framework, indicating a supervised fine-tuning (SFT) approach. The training process leveraged specific versions of popular ML frameworks, including TRL 0.17.0, Transformers 4.52.3, and PyTorch 2.5.1.

Good For

  • Applications requiring efficient, direct answers to multiple-choice questions.
  • Tasks where the answer is present in the context and does not necessitate advanced reasoning capabilities.
  • Integration into systems needing a compact yet capable MCQA model.