Environment captured by CLI; community evaluation suite attached.
5 reproductions
Use the exact revision and inspect the manifest before execution.
modelverse run --model hf:community/nova-instruct-12b@f13d2c0 --runtime vllmThe comparison context that makes the result interpretable.
Version 1.0 of the Apache-2.0 licensed Modelverse Run schema.
{
"schemaVersion": "1.0",
"id": "run_nova_3090_multilingual",
"createdAt": "2026-07-15T16:00:00.000Z",
"model": {
"provider": "huggingface",
"repository": "community/nova-instruct-12b",
"revision": "f13d2c0",
"quantization": "Q4_K_M",
"licenseSnapshot": "Apache-2.0"
},
"runtime": {
"name": "vLLM",
"version": "0.11.0",
"command": "modelverse run --model hf:community/nova-instruct-12b@f13d2c0 --runtime vllm"
},
"hardware": {
"os": "Ubuntu 26.04",
"architecture": "x64",
"cpu": "AMD Threadripper 3960X",
"systemMemoryMb": 131072,
"accelerators": [
{
"vendor": "NVIDIA",
"model": "NVIDIA GeForce RTX 3090",
"count": 1,
"memoryMb": 24576,
"driverVersion": "590.44",
"computeApi": "cuda"
}
]
},
"task": {
"type": "translation",
"evaluationSuite": "community-multilingual-instruction-v3",
"datasetRevision": "77dc012"
},
"configuration": {
"contextLength": 8192,
"maxGeneratedTokens": 512,
"temperature": 0.1,
"seed": 8,
"batchSize": 8,
"quantization": "Q4_K_M"
},
"metrics": {
"timeToFirstTokenMs": 178,
"tokensPerSecond": 118.4,
"peakMemoryMb": 19251.2,
"peakVramMb": 14028.8,
"scores": {
"Multilingual instruction suite": 71.3
}
},
"evidence": {
"machineCaptured": true,
"redactionsApplied": [],
"artifactUrls": []
}
}Batch throughput result; compare only with equivalent concurrency.