{ "format": "LLM-SCAN", "format_version": "1.0", "generator": { "name": "llmscan-engine", "version": "2.0.0" }, "scan_context": { "started_at": "2026-08-05T12:13:57+00:00Z", "ended_at": "2026-08-05T12:13:57+00:00Z", "duration_ms": 19, "limits": { "timeout_ms": 60000, "sandbox": true }, "mode": "static-deep" }, "source": { "input_type": "file", "filename": "nemo128.onnx", "size_bytes": 139764, "hashes": { "sha256": "a9fde1486ebfcc08f328d75ad4610c67835fea58c73ba57e3209a6f6cf019e9f" } }, "model": { "format": "onnx", "arch": null, "quantization": null, "params_estimate": 33426, "file_size_bytes": 139764, "sha256": "a9fde1486ebfcc08f328d75ad4610c67835fea58c73ba57e3209a6f6cf019e9f", "sha512": "cde05a330bf70e5fa87e8487ea7a83c55d6de8c3973ec8d2d03abece461510174f04de6f9b61bcc0109c0a5e6b506c7cd18020f0d317e6eb50971892534a8dd9", "kv_meta": {}, "scanner_model_info": { "producer_name": "OnnxScript", "producer_version": "0.3.2", "domain": "", "model_version": 0, "doc_string": "LogMelSpectrogram feature extractor for Nemo models", "ir_version": 8, "opsets": { "ai.onnx": 17 }, "node_count": 34, "unique_op_types": 20, "op_histogram": { "Div": 4, "Slice": 3, "Sub": 3, "Cast": 3, "Add": 3, "Where": 3, "ReduceSumSquare": 2, "Mul": 1, "Concat": 1, "Pad": 1, "STFT": 1, "MatMul": 1, "Log": 1, "Transpose": 1, "Unsqueeze": 1, "Shape": 1, "Range": 1, "Less": 1, "ReduceSum": 1, "Sqrt": 1 }, "op_risk_report": { "rows": [ { "op": "Div", "count": 4, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Slice", "count": 3, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Sub", "count": 3, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Cast", "count": 3, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Add", "count": 3, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Where", "count": 3, "risk": "normal", "note": "Standard ONNX op" }, { "op": "ReduceSumSquare", "count": 2, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Mul", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Concat", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Pad", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "STFT", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "MatMul", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Log", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Transpose", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Unsqueeze", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Shape", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Range", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Less", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "ReduceSum", "count": 1, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Sqrt", "count": 1, "risk": "normal", "note": "Standard ONNX op" } ], "elevated_categories": [], "has_control_flow": false, "has_memory_amp": false, "has_provider_compat": false, "has_runtime_sensitive": false, "has_quantized": false, "has_nondeterministic": false, "has_string_ops": false }, "inputs": [ { "name": "waveforms", "dtype": "FLOAT", "shape": [ "batch_size", "N" ] }, { "name": "waveforms_lens", "dtype": "INT64", "shape": [ "batch_size" ] } ], "outputs": [ { "name": "features", "dtype": "FLOAT", "shape": [ "batch_size", 128, "T" ] }, { "name": "features_lens", "dtype": "INT64", "shape": [ "batch_size" ] } ], "shape_risk_report": { "inputs": [ { "name": "waveforms", "dtype": "FLOAT", "shape": [ "batch_size", "N" ], "has_dynamic": true, "has_negative": false, "est_elements": 1 }, { "name": "waveforms_lens", "dtype": "INT64", "shape": [ "batch_size" ], "has_dynamic": true, "has_negative": false, "est_elements": 1 } ], "outputs": [ { "name": "features", "dtype": "FLOAT", "shape": [ "batch_size", 128, "T" ], "has_dynamic": true, "has_negative": false, "est_elements": 128 }, { "name": "features_lens", "dtype": "INT64", "shape": [ "batch_size" ], "has_dynamic": true, "has_negative": false, "est_elements": 1 } ], "any_dynamic_dims": true, "dynamic_dim_names": [ "N", "T", "batch_size" ], "negative_dims_detected": false, "extreme_dims_detected": false, "input_elements_est": 2, "output_elements_est": 129, "amplification_ratio": 64.5 }, "tensor_count": 16, "param_estimate": 33426, "weights_size_bytes": 135804, "weights_size_mb": 0.13, "dtype_histogram": { "FLOAT": 32901, "INT64": 13, "DOUBLE": 512 }, "custom_domains": [], "custom_domain_detail": {}, "custom_ops_report": { "has_non_standard": false, "standard_only": true, "domains": [] }, "has_external_data": false, "external_refs": [], "external_data_checks": { "present": false, "tensor_count": 0, "absolute_paths": [], "traversal_detected": [], "windows_drive_paths": [], "remote_url_paths": [], "null_byte_paths": [], "duplicate_files": [], "missing_files": [], "oob_reads": [], "size_validated_count": 0, "size_validation_total": 0 }, "quantized": false, "model_type": "audio-speech", "metadata_props": { "model_author": "Ilya Stupakov", "model_license": "MIT License" }, "inline_tensor_mismatches": 0, "zero_dim_initializers": 0, "extreme_dim_tensors": 0, "initializer_integrity": { "count": 16, "inline_tensor_mismatches": 0, "empty_initializers": 0, "duplicate_names": 0, "duplicate_name_list": [], "unused_by_graph": 0, "unused_names": [], "shadowing_graph_inputs": 0, "shadowing_names": [], "nan_inf_scan": { "performed": true, "method": "full", "tensors_scanned": 7, "total_float_tensors": 7, "values_scanned": 33413, "nan_count": 0, "inf_count": 0, "affected_tensors": [] } }, "declared_weight_bytes": 135804, "declared_weight_mb": 0.13, "weight_delta_bytes": 3960, "weight_delta_mb": 0.0, "file_size_bytes": 139764, "file_size_mb": 0.13, "sha256": "a9fde1486ebfcc08f328d75ad4610c67835fea58c73ba57e3209a6f6cf019e9f", "sha512": "cde05a330bf70e5fa87e8487ea7a83c55d6de8c3973ec8d2d03abece461510174f04de6f9b61bcc0109c0a5e6b506c7cd18020f0d317e6eb50971892534a8dd9", "dtype_stats": { "FLOAT": { "count": 6, "bytes": 131604, "mb": 0.13, "avg_bits_per_elem": 32.0 }, "INT64": { "count": 9, "bytes": 104, "mb": 0.0, "avg_bits_per_elem": 64.0 }, "DOUBLE": { "count": 1, "bytes": 4096, "mb": 0.0, "avg_bits_per_elem": 64.0 } }, "tensor_preview": [ { "index": 0, "name": "tmp", "dims": [ 257, 128 ], "dtype_name": "FLOAT", "byte_len": 131584, "bits_per_elem": 32.0 }, { "index": 1, "name": "int64_1_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 2, "name": "int64_0_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 3, "name": "int64_9223372036854775807_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 4, "name": "int64_m1_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 5, "name": "preemph_2_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 6, "name": "tmp_11", "dims": [ 4 ], "dtype_name": "INT64", "byte_len": 32, "bits_per_elem": 64.0 }, { "index": 7, "name": "hann_window", "dims": [ 512 ], "dtype_name": "DOUBLE", "byte_len": 4096, "bits_per_elem": 64.0 }, { "index": 8, "name": "hop_length", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 9, "name": "log_zero_guard_value_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 10, "name": "int64_1_cast", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 11, "name": "const_2", "dims": [ 2 ], "dtype_name": "INT64", "byte_len": 16, "bits_per_elem": 64.0 }, { "index": 12, "name": "int64_0_cast", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 13, "name": "int64_0_2_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 14, "name": "int64_1_9_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 15, "name": "const_13_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 } ], "op_histogram_full": { "Div": 4, "Slice": 3, "Sub": 3, "Cast": 3, "Add": 3, "Where": 3, "ReduceSumSquare": 2, "Mul": 1, "Concat": 1, "Pad": 1, "STFT": 1, "MatMul": 1, "Log": 1, "Transpose": 1, "Unsqueeze": 1, "Shape": 1, "Range": 1, "Less": 1, "ReduceSum": 1, "Sqrt": 1 } }, "opsets": { "ai.onnx": 17 }, "producer": "OnnxScript", "domain": "", "ir_version": 8, "nodes": 34, "initialized_tensors": 16, "weights_size_est_bytes": 135804, "inputs": [ { "name": "waveforms", "dtype": "FLOAT", "shape": [ "batch_size", "N" ] }, { "name": "waveforms_lens", "dtype": "INT64", "shape": [ "batch_size" ] } ], "outputs": [ { "name": "features", "dtype": "FLOAT", "shape": [ "batch_size", 128, "T" ] }, { "name": "features_lens", "dtype": "INT64", "shape": [ "batch_size" ] } ], "op_hist_top": { "Div": 4, "Slice": 3, "Sub": 3, "Cast": 3, "Add": 3, "Where": 3, "ReduceSumSquare": 2, "Mul": 1, "Concat": 1, "Pad": 1, "STFT": 1, "MatMul": 1, "Log": 1, "Transpose": 1, "Unsqueeze": 1, "Shape": 1, "Range": 1, "Less": 1, "ReduceSum": 1, "Sqrt": 1 }, "dtype_hist": { "FLOAT": 32901, "INT64": 13, "DOUBLE": 512 }, "custom_domains": [], "has_external_data": false, "external_refs": [], "quantized": false, "model_type": "audio-speech", "file_size_mb": 0.13, "declared_weight_bytes": 135804, "declared_weight_mb": 0.13, "weight_delta_mb": 0.0, "dtype_stats": { "FLOAT": { "count": 6, "bytes": 131604, "mb": 0.13, "avg_bits_per_elem": 32.0 }, "INT64": { "count": 9, "bytes": 104, "mb": 0.0, "avg_bits_per_elem": 64.0 }, "DOUBLE": { "count": 1, "bytes": 4096, "mb": 0.0, "avg_bits_per_elem": 64.0 } }, "tensor_preview": [ { "index": 0, "name": "tmp", "dims": [ 257, 128 ], "dtype_name": "FLOAT", "byte_len": 131584, "bits_per_elem": 32.0 }, { "index": 1, "name": "int64_1_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 2, "name": "int64_0_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 3, "name": "int64_9223372036854775807_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 4, "name": "int64_m1_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 5, "name": "preemph_2_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 6, "name": "tmp_11", "dims": [ 4 ], "dtype_name": "INT64", "byte_len": 32, "bits_per_elem": 64.0 }, { "index": 7, "name": "hann_window", "dims": [ 512 ], "dtype_name": "DOUBLE", "byte_len": 4096, "bits_per_elem": 64.0 }, { "index": 8, "name": "hop_length", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 9, "name": "log_zero_guard_value_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 10, "name": "int64_1_cast", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 11, "name": "const_2", "dims": [ 2 ], "dtype_name": "INT64", "byte_len": 16, "bits_per_elem": 64.0 }, { "index": 12, "name": "int64_0_cast", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 13, "name": "int64_0_2_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 14, "name": "int64_1_9_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 15, "name": "const_13_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 } ] }, "tokenizer": {}, "tensors": { "count": 16, "preview": [ { "index": 0, "name": "tmp", "dims": [ 257, 128 ], "dtype_name": "FLOAT", "byte_len": 131584, "bits_per_elem": 32.0 }, { "index": 1, "name": "int64_1_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 2, "name": "int64_0_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 3, "name": "int64_9223372036854775807_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 4, "name": "int64_m1_1d", "dims": [ 1 ], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 5, "name": "preemph_2_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 6, "name": "tmp_11", "dims": [ 4 ], "dtype_name": "INT64", "byte_len": 32, "bits_per_elem": 64.0 }, { "index": 7, "name": "hann_window", "dims": [ 512 ], "dtype_name": "DOUBLE", "byte_len": 4096, "bits_per_elem": 64.0 }, { "index": 8, "name": "hop_length", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 9, "name": "log_zero_guard_value_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 10, "name": "int64_1_cast", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 11, "name": "const_2", "dims": [ 2 ], "dtype_name": "INT64", "byte_len": 16, "bits_per_elem": 64.0 }, { "index": 12, "name": "int64_0_cast", "dims": [], "dtype_name": "INT64", "byte_len": 8, "bits_per_elem": 64.0 }, { "index": 13, "name": "int64_0_2_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 14, "name": "int64_1_9_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 }, { "index": 15, "name": "const_13_cast", "dims": [], "dtype_name": "FLOAT", "byte_len": 4, "bits_per_elem": 32.0 } ], "dtype_histogram": { "FLOAT": 32901, "INT64": 13, "DOUBLE": 512 }, "dtype_stats": { "FLOAT": { "count": 6, "bytes": 131604, "mb": 0.13, "avg_bits_per_elem": 32.0 }, "INT64": { "count": 9, "bytes": 104, "mb": 0.0, "avg_bits_per_elem": 64.0 }, "DOUBLE": { "count": 1, "bytes": 4096, "mb": 0.0, "avg_bits_per_elem": 64.0 } }, "integrity": { "declared_weight_bytes": 135804, "weights_size_est_bytes": 135804, "weight_delta_mb": 0.0, "has_external_data": false, "external_refs": [] } }, "security": { "grade": "B", "score": 88, "findings": [ { "severity": "INFO", "category": "Shape Risk", "title": "Dynamic tensor dimensions detected", "detail": "Inputs with dynamic dims: ['waveforms', 'waveforms_lens']. Outputs: ['features', 'features_lens'] (symbolic: N, T, batch_size). Dynamic dims allow flexible batch sizes and sequence lengths, but without enforced bounds a caller can supply arbitrarily large tensors causing OOM. This is expected for most inference models.", "recommendation": "Enforce maximum input shape at the serving layer (e.g. ORT SessionOptions.add_session_config_entry, preprocessing validation). Document accepted input bounds in model card or deployment README." }, { "severity": "LOW", "category": "Shape Risk", "title": "High output/input element ratio: 64.5×", "detail": "Estimated output elements (129) are 64.5× larger than input elements (2). Large amplification ratios can indicate detection heads with many anchors, segmentation masks, or other outputs that scale poorly with batch size — relevant for memory planning in production.", "recommendation": "Profile inference memory at maximum expected batch size before production deployment." }, { "severity": "LOW", "category": "Supply Chain", "title": "No license field in ONNX model metadata", "detail": "Models without license information cannot be safely redistributed or deployed commercially. ONNX metadata supports a license property in model.metadata_props.", "recommendation": "Add a license field during export (e.g. apache-2.0, mit, cc-by-4.0)" }, { "severity": "LOW", "category": "Supply Chain", "title": "No provenance metadata in ONNX model", "detail": "Neither author nor domain (producer namespace) are set. Without origin information the model cannot be traced to a trusted source.", "recommendation": "Set author and domain fields during model export" }, { "severity": "INFO", "category": "Provenance", "title": "Model producer: OnnxScript 0.3.2", "detail": "Producer metadata helps identify the tool that exported this model", "recommendation": "Verify producer matches expected export pipeline" }, { "severity": "INFO", "category": "Classification", "title": "Inferred model type: audio-speech", "detail": "Based on op inventory (34 total nodes, 20 unique types)", "recommendation": "Verify this matches the model's intended use case" }, { "severity": "INFO", "category": "Runtime", "title": "Default opset: 17", "detail": "Opset governs which ops and behaviors are available at runtime", "recommendation": "Use ONNX Runtime (CPU/CUDA/DirectML/OpenVINO backends); requires ORT ≥ 1.14" }, { "severity": "LOW", "category": "Runtime Safety", "title": "Fully dynamic input shapes (2 input(s))", "detail": "Inputs with all-dynamic dimensions: [('waveforms', \"['batch_size', 'N']\"), ('waveforms_lens', \"['batch_size']\")]. Without static bounds, a malicious caller can supply arbitrarily large tensors causing OOM. The model itself is not malicious, but deployments must enforce input size limits.", "recommendation": "Enforce maximum input shape at the serving layer (e.g. ORT session options, preprocessing validation). Document accepted input bounds in model card." } ] }, "compatibility": [ { "runtime": "ONNX Runtime", "direct_supported": true, "conversion_possible": true, "verification_required": false, "quant": null, "arch": true, "context": "opset 17 (requires ORT ≥ 1.14)", "notes": "ONNX Runtime (CPU/CUDA/DirectML/OpenVINO backends); requires ORT ≥ 1.14" }, { "runtime": "TensorRT", "direct_supported": false, "conversion_possible": true, "verification_required": true, "quant": null, "arch": null, "context": "opset 17; dynamic batch: needs profiles", "notes": "Engine build required (trtexec / TRT Python API) | dynamic batch requires explicit optimization profiles (min/opt/max shapes)" }, { "runtime": "OpenVINO", "direct_supported": false, "conversion_possible": true, "verification_required": true, "quant": null, "arch": null, "context": "opset 17; conversion required", "notes": "Conversion via Model Optimizer (mo) or OpenVINO OVC (openvino.convert_model) required | Op coverage varies; validate with mo --input_model before deployment | dynamic shapes supported but may require --input re-specification" }, { "runtime": "CoreML (Apple)", "direct_supported": false, "conversion_possible": false, "verification_required": true, "quant": null, "arch": null, "context": "opset 17; conversion via coremltools; opset out of range", "notes": "Conversion path via coremltools.converters.onnx.convert() — CoreML is not a direct ONNX runtime; the model must be converted to .mlpackage first. opset 17 exceeds CoreML converter limit (≤ 16). " }, { "runtime": "llama.cpp / Ollama", "direct_supported": false, "conversion_possible": false, "verification_required": false, "quant": false, "arch": null, "context": null, "not_supported_msg": "ONNX not supported — requires GGUF format", "notes": "GGUF format only; convert with llama.cpp convert scripts" } ], "report": { "summary": "ONNX audio-speech model | opset 17 | 34 nodes | 33,426 params | 4 warnings / 4 informational", "bullets": [ "ONNX audio-speech model | opset 17 | 34 nodes | 33,426 params | 4 warnings / 4 informational", "Size: 0.13 MB", "SHA256: a9fde1486ebfcc08..." ], "recommendations": [ "Profile inference memory at maximum expected batch size before production deployment.", "Add a license field during export (e.g. apache-2.0, mit, cc-by-4.0)", "Set author and domain fields during model export", "Enforce maximum input shape at the serving layer (e.g. ORT session options, preprocessing validation). Document accepted input bounds in model card." ] } }