Vela-2.0-0.3B

The previous default decision model, a ModernBERT encoder for a CPU. It answers set and span questions as well, with calibrated thresholds for the built-in questions.

Property	Value
Repository	vllm-sr/Vela-2.0-0.3B, pinned at a3209a50dc3e
Family	Vela 2.0
Parameters	0.309B loaded
Backbone	ModernBERT
Precision (--dtype auto)	FP32 on every device
Weights	about 1.2 GB at FP32
Device memory	4 GiB or more
Reads	text

Question types

Type	Answers
choice	Yes
noul	Yes
score	Yes
set	Yes
span	Yes, on the router span head

Built-in questions

The Vela 2.0 family was trained on the Router's built-in questions, so the Router asks each one in its trained form; the support matrix lists which.

A rule without a predicate takes this model's calibrated threshold:

Preset	Threshold
jailbreak (jailbreak)	0.75
domain (domain)	0.28
pii (pii)	0.01
fact check (fact_check)	0.93
user feedback (feedback)	0.37
safety (harm)	0.46
modality (modality)	0.51

Hardware

Each time it loads, the model compares its answers with the reference answers recorded for its revision. They are recorded for:

• CPU (cpu)
• NVIDIA Hopper (sm_90) (cuda:sm_90)
• Ascend NPU (npu)
• AMD GPU (ROCm) (rocm)

Where a hardware reference and its device class's are both recorded, the hardware one is checked.

Choose a model lists the validated hardware.

Quantization

The runtime loads no quantized packages of the Vela 2.0 family: it runs at its published precision. Batches that merge requests run a pre-packed FP32 copy of its linear layers on x86 CPUs with oneDNN, which its records show changes no decision.

Deploy it

Serve it on its own in Engine mode, to ask it directly:

CPU

vllm-sr serve vllm-sr/Vela-2.0-0.3B --engine --platform cpu

AMD GPU

vllm-sr serve vllm-sr/Vela-2.0-0.3B --engine --platform rocm

NVIDIA GPU

vllm-sr serve vllm-sr/Vela-2.0-0.3B --engine --platform cuda

Or make it the Router's decision model:

config.yaml

global:
  model_catalog:
    deployments:
      primary:
        provider: model_runtime
        artifact: vllm-sr/Vela-2.0-0.3B
        device: auto
    system:
      decision_model: {deployment: primary}

device: auto takes the first validated GPU with enough free memory, else the CPU; name rocm:0 or cuda:0 to pin a GPU.

To run it on another host and attach the Router to it:

vllm-srun serve vllm-sr/Vela-2.0-0.3B --device auto --host 0.0.0.0 --port 8100

Then set endpoint: http://<host>:8100 and model: Vela-2.0-0.3B on the deployment; see Attach to a runtime you run.

Learn more

• Model card on Hugging Face
• vela2-router-signals record
• vela2-parity record
• vela2-performance record
• router-latency-cpu record
• Vela 2.0: the family's sizes and how they differ