Overview

DeepScaleR fine-tuned DeepSeek-R1-Distill-Qwen-1.5B with RL while gradually extending the context length, and released the data, code and logs.

Highlights

  • 1.5B parameters
  • RL-trained for math reasoning
  • Open training recipe
  • MIT license

Typical uses

Math reasoning on very small hardware and RL research.

Things to consider

  • Narrow focus on math.
  • Download sizes refer to the default quantized versions in the Ollama library; full-precision weights are larger.

Specifications are taken from the model card on Hugging Face, the Ollama library and the provider’s announcement. Benchmark results depend on evaluation settings; see how AI benchmarks work.