Skip to content

decima:agent

4 TagsUpdated 321M params2048 context100+ languagesApache-2.0by A. M. Madani

Decima-agent 2.1: decima-base fine-tuned for coding-agent decisions (secret and command gates, tool and model choice); states up to 2,048 tokens.

multilingual321m
ollaya run decima:agent --preset triage "I was charged twice for my subscription this month and want a refund."

Details

  • graph713a903f75c2 · 5 MBonnx · decima · 321M · fp32
  • weightsbd47dc3f80bf · 1.2 GBhuggingface.co/amyrmahdy/decima-agent/resolve/86a07aa…/pytorch/encoder/model.safetensors
  • weights5d865ca48f72 · 57 MBhuggingface.co/amyrmahdy/decima-agent/resolve/86a07aa…/pytorch/head.safetensors
  • tokenizerd941237f7b66 · 34 MBhuggingface.co/amyrmahdy/decima-agent/resolve/86a07aa…/pytorch/encoder/tokenizer.json
  • decision6deba326a211 · 3 KB{"engine": "onnx", "family": "decima", "encoder": "", "layout": "decima-late-interaction-v1", …}
  • calibrationd94b8180c1b8 · 291 B{"temperature": [0.608, 1.0, 0.608]}
  • 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