Skip to content

winnow:e2b

6 TagsUpdated 4.6B params8192 context100+ languagesApache-2.0by EldanRing

Winnow-E2B, Q8_0 GGUF (5 GB), uncalibrated: the smallest Winnow, 0.616 on typed decisions in 104 ms on an RTX 5090.

multilingual4.6b
ollaya run winnow:e2b --preset triage "I was charged twice for my subscription this month and want a refund."

Details

  • weightscb37414b46b5 · 5.0 GBgguf · Q8_0 · huggingface.co/EldanRing/Winnow-E2B/resolve/00b4d4e…/gguf/Winnow-E2B-Q8_0.gguf
  • decisiona5d976250892 · 2 KB{"engine": "llama", "family": "winnow", "layout": "winnow-v1", "gguf": {…}, …}
  • calibration42fa34de79f2 · 130 B{"temperature": [1.0, 1.0, 1.0]}
  • license9c37c6e87ed9 · 11 KBWinnow-E2B

Every layer is checked against its sha256 when it is pulled. Weights and tokenizers download from the model author's Hugging Face repository at a pinned commit; Ollaya never re-hosts them.

Winnow readme and all models