SPX-CD-Omni API & Inference Endpoint

Download

bash

pip install huggingface_hubhf download SurdAI/SPX-CD-Omni --local-dir SPX-CD-Omnicd SPX-CD-Omni

Run inference

Install CUDA-enabled PyTorch for Transformers. Use separate environments for the two backends.

Transformers — BF16 base:

bash

pip install -r requirements-transformers.txtpython infer.py --backend transformers --adapter . \  --input examples/text.json --output predictions.jsonl

vLLM — AWQ INT4 base:

bash

pip install -r requirements-vllm.txtpython infer.py --backend vllm --base nicklas373/gemma-4-12B-it-AWQ \  --adapter vllm-adapter --input examples/text.json --output predictions.jsonl

For images or multi-select, use examples/image.json or examples/multi_select.json as the input. See these files for the input format and python infer.py --help for options.

Run the audio test

bash

pip install -r requirements-audio.txtpython smoke_audio.py --adapter vllm-adapter --output audio-results.json

Tests four bundled speech clips and silent controls. Base and adapter each scored 3/4.

Results

Self-evaluated with the AWQ INT4 base.

Table with columns: Benchmark / subset, Effort, Questions, Accuracy ↑, I ↑, C ↑, ECE ↓, TVD ↓, RPS ↓
Benchmark / subset	Effort	Questions	Accuracy ↑	I ↑	C ↑	ECE ↓	TVD ↓	RPS ↓
CMDB-1500	1	1,500	73.00%	45.77	77.38	0.0246	0.4018	0.1205
CMDB text	1	1,200	71.00%	45.63	75.75	0.0220	0.5013	0.1205
CMDB images (native)	1	300	81.00%	70.43

JevBench ECE is Choice / Noul / Score. N/A means no rating questions. Metric scopes and sample counts are in METRICS.md; reproduction commands are in EVALUATION.md.

Settings

--effort 1–5: average that many distinct candidate orderings, capped at two for binary questions. Reported results cover efforts 1 and 2.
--temperature: candidate-softmax temperature, default 1.0.
--prompt-format: cmdb for CMDB; open-format for JevBench and Decision Bench.

Base: Google Gemma 4 12B, Apache 2.0.

Download

bash

pip install huggingface_hubhf download SurdAI/SPX-CD-Omni --local-dir SPX-CD-Omnicd SPX-CD-Omni

Run inference

Install CUDA-enabled PyTorch for Transformers. Use separate environments for the two backends.

Transformers — BF16 base:

bash

pip install -r requirements-transformers.txtpython infer.py --backend transformers --adapter . \  --input examples/text.json --output predictions.jsonl

vLLM — AWQ INT4 base:

bash

pip install -r requirements-vllm.txtpython infer.py --backend vllm --base nicklas373/gemma-4-12B-it-AWQ \  --adapter vllm-adapter --input examples/text.json --output predictions.jsonl

For images or multi-select, use examples/image.json or examples/multi_select.json as the input. See these files for the input format and python infer.py --help for options.

Run the audio test

bash

pip install -r requirements-audio.txtpython smoke_audio.py --adapter vllm-adapter --output audio-results.json

Tests four bundled speech clips and silent controls. Base and adapter each scored 3/4.

Results

Self-evaluated with the AWQ INT4 base.

Table with columns: Benchmark / subset, Effort, Questions, Accuracy ↑, I ↑, C ↑, ECE ↓, TVD ↓, RPS ↓
Benchmark / subset	Effort	Questions	Accuracy ↑	I ↑	C ↑	ECE ↓	TVD ↓	RPS ↓
CMDB-1500	1	1,500	73.00%	45.77	77.38	0.0246	0.4018	0.1205
CMDB text	1	1,200	71.00%	45.63	75.75	0.0220	0.5013	0.1205
CMDB images (native)	1	300	81.00%	70.43

JevBench ECE is Choice / Noul / Score. N/A means no rating questions. Metric scopes and sample counts are in METRICS.md; reproduction commands are in EVALUATION.md.

Settings

--effort 1–5: average that many distinct candidate orderings, capped at two for binary questions. Reported results cover efforts 1 and 2.
--temperature: candidate-softmax temperature, default 1.0.
--prompt-format: cmdb for CMDB; open-format for JevBench and Decision Bench.

Base: Google Gemma 4 12B, Apache 2.0.

SPX-CD-Omni

README

Download

Run inference

Run the audio test

Results

Settings

Explore FriendliAI today

README

Download

Run inference

Run the audio test

Results

Settings