HunyuanOCR-mic / integration.json
tarekziade's picture
tarekziade HF Staff
Measure checkpoints above 1B on GPU in bfloat16 only
2b69a85 verified
Raw History Blame Contribute Delete
2.13 kB
{
"mic": "0.3",
"name": "HunyuanOCR document parsing",
"usage": "AutoProcessor + HunYuanVLForConditionalGeneration, chat template with image + text, batched with padding, inference probe and one SGD step on the language-model head",
"architecture": "HunYuanVLForConditionalGeneration",
"checkpoints": {
"tencent/HunyuanOCR": {
"fixture": "huggingface-mic/HunyuanOCR-mic"
}
},
"fixtures": {
"huggingface-mic/HunyuanOCR-mic": {
"built_from": "tencent/HunyuanOCR",
"covers": [
"tencent/HunyuanOCR"
],
"build": {
"seed": 17,
"model_class": "HunYuanVLForConditionalGeneration"
}
}
},
"integrations": {
"transformers": {
"entry_point": "integrations/transformers/usage.md",
"format": "runnable-markdown",
"requires": [
"transformers>=5",
"torch",
"torchvision"
],
"targets": [
"fixtures",
"checkpoints"
],
"variants": {
"default": {},
"eager": {
"attn_implementation": "eager"
},
"bfloat16": {
"dtype": "bfloat16"
}
},
"training": true,
"resources": {
"fixture": {
"memory_gb": 4,
"download_mb": 40
},
"tencent/HunyuanOCR": {
"memory_gb": 32,
"gpu_memory_gb": 20,
"download_mb": 4000,
"variants": {
"bfloat16": {
"gpu_memory_gb": 16
},
"default": {
"devices": []
},
"eager": {
"devices": []
}
},
"devices": [
"cuda"
]
}
}
}
},
"outputs": {
"runtime": {
"compare": "exact"
},
"logits.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"
}
}
}