tarekziade's picture
tarekziade HF Staff
Publish MIC usage, slim fixture and reviewed Transformers expectations
de5f6eb verified
Raw History Blame Contribute Delete
2.26 kB
{
"mic": "0.3",
"name": "OWLv2 open-vocabulary object detection",
"usage": "Owlv2Processor + Owlv2ForObjectDetection, two synthetic images, two text queries, post-processing, and one SGD step on detection labels",
"architecture": "Owlv2ForObjectDetection",
"checkpoints": {
"google/owlv2-base-patch16-ensemble": {
"fixture": "huggingface-mic/owlv2-base-patch16-ensemble-mic"
}
},
"fixtures": {
"huggingface-mic/owlv2-base-patch16-ensemble-mic": {
"built_from": "google/owlv2-base-patch16-ensemble",
"covers": [
"google/owlv2-base-patch16-ensemble"
],
"build": {
"seed": 17,
"model_class": "Owlv2ForObjectDetection",
"accept_insensitive": {
"attention_mask:all_ones": "OWLv2's text tower is causal and pools at the EOS token, so padding after EOS cannot move any output"
}
}
}
},
"integrations": {
"transformers": {
"entry_point": "integrations/transformers/usage.md",
"format": "runnable-markdown",
"requires": [
"transformers>=5",
"torch",
"torchvision"
],
"targets": [
"fixtures",
"checkpoints"
],
"variants": {
"default": {},
"bfloat16": {
"dtype": "bfloat16"
}
},
"training": true,
"resources": {
"fixture": {
"memory_gb": 4,
"download_mb": 50
},
"google/owlv2-base-patch16-ensemble": {
"memory_gb": 8,
"gpu_memory_gb": 8,
"download_mb": 1200,
"variants": {
"bfloat16": {
"gpu_memory_gb": 6
}
}
}
}
}
},
"outputs": {
"runtime": {
"compare": "exact"
},
"detections": {
"compare": "exact"
},
"logits_probe": {
"compare": "allclose",
"atol": 0.0001,
"rtol": 0.0001
},
"boxes_probe": {
"compare": "allclose",
"atol": 0.0001,
"rtol": 0.0001
},
"training.loss": {
"compare": "allclose",
"atol": 0.0001,
"rtol": 0.0001
},
"training.gradients": {
"compare": "all_true"
},
"training.updated": {
"compare": "true"
}
}
}