Skip to content

decima:guard-small

6 TagsUpdated 140M params2048 context100+ languagesApache-2.0by A. M. Madani

Decima-guard-small 1.0 preview (mmBERT-small, 140M): the same checks, faster, a few points lower.

multilingual140m
ollaya run decima:guard-small --preset triage "I was charged twice for my subscription this month and want a refund."

Details

  • graphf2cdf2881351 · 5 MBonnx · decima · 140M · fp32
  • weights6e8efdaee4c1 · 562 MBhuggingface.co/amyrmahdy/decima-guard-small/resolve/ac11024…/pytorch/encoder/model.safetensors
  • weights8c5fbbfec97a · 19 MBhuggingface.co/amyrmahdy/decima-guard-small/resolve/ac11024…/pytorch/head.safetensors
  • tokenizerd941237f7b66 · 34 MBhuggingface.co/amyrmahdy/decima-guard-small/resolve/ac11024…/pytorch/encoder/tokenizer.json
  • decision2663dee698bc · 3 KB{"engine": "onnx", "family": "decima", "encoder": "", "layout": "decima-late-interaction-v2", …}
  • calibration45d523702d77 · 291 B{"temperature": [0.755, 1.0, 0.755]}
  • license6e583cb7b1fd · 11 KBDecima-small, Decima-base, Decima-agent, Decima-guard and Decima-guard-small 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