reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT

Hugging Face
TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.8BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Mar 22, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Warm

The reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT is a 0.6 billion parameter Qwen3-based causal language model developed by Convergent Intelligence LLC: Research Division. It was created through a two-stage distillation process: first from a 30B-parameter "Thinking" teacher for STEM reasoning, then fine-tuned on legal instruction data. This model is optimized for ultra-lightweight reasoning and legal instruction-following, designed for deployment on resource-constrained devices like mobile phones, offering a 50x compression ratio from its teacher model.

Loading preview...

Overview

reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT is a compact 0.6 billion parameter Qwen3-based model developed by Convergent Intelligence LLC: Research Division. It is engineered for efficient reasoning and legal instruction-following, achieved through a unique two-stage training pipeline. The model undergoes a 50x compression from its 30B-parameter teacher, making it suitable for deployment on devices with limited resources, such as mobile phones, with quantized versions under 500MB.

Key Capabilities

  • Structured Reasoning: Distilled from a "Thinking" teacher model, it establishes a robust backbone for structured reasoning, particularly in STEM domains.
  • Legal Instruction-Following: Fine-tuned on legal instruction data, enabling it to handle legal analysis and queries.
  • Ultra-Lightweight Deployment: Designed for edge and mobile devices, offering significant parameter compression while retaining reasoning capabilities.
  • Derivation-Focused Training: Utilizes proof-weighted cross-entropy during distillation, emphasizing reasoning steps over final answers.

Good For

  • Ultra-lightweight reasoning: Ideal for mobile, edge, and IoT applications requiring reasoning capabilities in a small footprint.
  • Legal and STEM instruction-following: Excels at tasks requiring structured derivation in STEM and legal analysis.
  • Educational tutoring: Can be used for generating explanations and derivations in STEM subjects.
  • Embedded inference: Suitable for scenarios where models need to run directly on devices with limited computational power.
  • Component in multi-model pipelines: Can serve as an efficient reasoning module within larger AI systems.

Finetunes
2 models