MMOPD/Qwen3-1.7B-OT3-2ep

TEXT GENERATIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:2BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 14, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

MMOPD/Qwen3-1.7B-OT3-2ep is a 1.7 billion parameter language model based on Qwen3-1.7B-Base, fine-tuned for two epochs on the OpenThoughts3-1.2M dataset. This model specializes in long-chain-of-thought reasoning across math, code, and science, designed to produce answers starting with a dedicated thinking block. With a context length of 32,768 tokens, it serves as a general reasoning student model within the MMOPD project for further distillation experiments and domain-specific teacher models. Its primary strength lies in structured reasoning and problem-solving.

Loading preview...

MMOPD/Qwen3-1.7B-OT3-2ep: A Reasoning-Focused Qwen3 Model

MMOPD/Qwen3-1.7B-OT3-2ep is a 1.7 billion parameter model derived from Qwen3-1.7B-Base, fine-tuned for two epochs on the extensive OpenThoughts3-1.2M dataset. This dataset comprises 1.2 million examples of long-chain-of-thought reasoning traces covering mathematics, coding, and scientific problems. The model is specifically engineered to be a "thinking" model, where every generated response begins with a <think> block, encouraging explicit reasoning steps.

Key Capabilities and Features

  • Structured Reasoning: Designed to articulate thought processes before providing a final answer, making its reasoning transparent.
  • Broad Reasoning Domains: Proficient in math, code, and science reasoning due to its specialized training data.
  • Extended Context Window: Supports a sequence length of up to 32,768 tokens, enabling complex problem-solving and understanding of lengthy prompts.
  • MMOPD Project Base: Serves as a foundational general reasoning student model for the MMOPD (multi-teacher on-policy distillation) project, used as a starting point for creating domain-specific teacher models.

Performance Highlights

Evaluations show its reasoning capabilities across various benchmarks, including AIME (25.0% on AIME24), LiveCodeBench v6 (19.7%), IFEval (30.9%), and IFBench (20.7%). These scores reflect its performance in complex reasoning tasks, particularly when compared to larger models in the series like Qwen3-4B-OT3-2ep.

When to Use This Model

This model is ideal for applications requiring explicit, step-by-step reasoning, particularly in educational tools, automated problem solvers, or as a base for further fine-tuning on specific reasoning-intensive tasks. Its unique thinking format makes it suitable for scenarios where understanding the 'how' behind an answer is as important as the answer itself.

Finetunes
1 models