oioxo-writer

A tiny on-device writer for oioxo โ€” produces clean summaries and short articles from source text, running entirely in the browser via transformers.js (no server, private).

  • Base: SmolLM2-135M-Instruct (Apache-2.0)
  • Trained by knowledge distillation from a Qwen2.5-7B-Instruct teacher (Apache-2.0) on Wikipedia-derived synthesis pairs.
  • onnx/model_q4.onnx โ€” 4-bit weights + int8 embeddings (~92 MB)
  • onnx/model_quantized.onnx โ€” int8 (~130 MB)

License

Apache-2.0 (inherits from the SmolLM2-135M base). Attribution to SmolLM2 retained.

Downloads last month
9
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for payam1394/oioxo-writer

Quantized
(143)
this model