Skip to content

decision2:kai

6 TagsUpdated 0.60B params8192 contextEnglishApache-2.0by vLLM Semantic Router

Qwen3-0.6B-Base, the fastest: 31 ms for five questions on an RTX 5090; 8k-token rows.

0.60b
ollaya run decision2:kai --preset triage "I was charged twice for my subscription this month and want a refund."

Details

  • graph4eb1e3d4919e · 6 MBonnx · decision2 · 0.60B · fp32
  • weights43369a2bfddb · 1.5 GBhuggingface.co/vllm-sr/Decision-2.0-Kai-0.6B/resolve/cc7f30c…/backbone/model.safetensors
  • weights508dfe385c8f · 4 MBhuggingface.co/vllm-sr/Decision-2.0-Kai-0.6B/resolve/cc7f30c…/decision_head.safetensors
  • tokenizerbe75606093db · 11 MBhuggingface.co/vllm-sr/Decision-2.0-Kai-0.6B/resolve/cc7f30c…/tokenizer.json
  • decision634fae7097f7 · 5 KB{"engine": "onnx", "family": "decision2", "encoder": "", "layout": "decision2-endpoint-v1", …}
  • calibrationbf756c46be45 · 187 B{"temperature": [1.0, 1.0, 1.0]}
  • license1314ad54d94f · 10 KBDecision 2.0 by the vLLM Semantic Router contributors (https://huggingface.co/vllm-sr)

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.

Decision 2.0 readme and all models