Skip to content

decima:small

4 TagsUpdated 122M params512 context100+ languagesApache-2.0by A. M. Madani

Decima-small 1.1 (multilingual-e5-small, 122M), fp32: 0.432 on typed decisions; 146 ms for five questions on a CPU, the fastest there.

multilingual122m
ollaya run decima:small --preset triage "I was charged twice for my subscription this month and want a refund."

Details

  • graphf83c59768e53 · 3 MBonnx · decima · 122M · fp32
  • weights2bc43026b30d · 471 MBhuggingface.co/amyrmahdy/decima-small/resolve/2e7f4d0…/pytorch/encoder/model.safetensors
  • weights5f2a149e4a4f · 19 MBhuggingface.co/amyrmahdy/decima-small/resolve/2e7f4d0…/pytorch/head.safetensors
  • tokenizer255f5e32cb32 · 17 MBhuggingface.co/amyrmahdy/decima-small/resolve/2e7f4d0…/pytorch/encoder/tokenizer.json
  • decisionf9f995481ae1 · 3 KB{"engine": "onnx", "family": "decima", "encoder": "", "layout": "decima-late-interaction-v1", …}
  • calibrationeb3ca5d61970 · 291 B{"temperature": [0.936, 1.0, 0.936]}
  • licensed2a0668879d4 · 11 KBDecima-small, Decima-base and Decima-agent by A. M. Madani (https://huggingface.co/amyrmahdy, https://github.com/amyrmahdy/decima), Apache-2.0.

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.

Decima readme and all models