{ "format": "LLM-SCAN", "format_version": "1.0", "generator": { "name": "llmscan-engine", "version": "2.0.0" }, "scan_context": { "started_at": "2026-08-05T12:17:14+00:00Z", "ended_at": "2026-08-05T12:17:14+00:00Z", "duration_ms": 3549, "limits": { "timeout_ms": 60000, "sandbox": true }, "mode": "static-deep" }, "source": { "input_type": "file", "filename": "encoder-model.int8.onnx", "size_bytes": 652183999, "hashes": { "sha256": "6139d2fa7e1b086097b277c7149725edbab89cc7c7ae64b23c741be4055aff09" } }, "model": { "format": "onnx", "arch": null, "quantization": null, "params_estimate": 608855720, "file_size_bytes": 652183999, "sha256": "6139d2fa7e1b086097b277c7149725edbab89cc7c7ae64b23c741be4055aff09", "sha512": "90d6b7163914efd49ad006071282404b788911b2b26f8a5fe70407f193bbce2194a90a92c9f63cec9b7c810146889deb443ffdaf9eb9b0a1cd5b592023ab6be5", "kv_meta": {}, "scanner_model_info": { "producer_name": "onnx.quantize", "producer_version": "0.1.0", "domain": "", "model_version": 0, "doc_string": null, "ir_version": 8, "opsets": { "ai.onnx": 17 }, "node_count": 5654, "unique_op_types": 34, "op_histogram": { "Constant": 1661, "Mul": 733, "Cast": 426, "Unsqueeze": 328, "Reshape": 295, "Transpose": 244, "DynamicQuantizeLinear": 223, "Concat": 221, "MatMulInteger": 217, "Add": 205, "Shape": 175, "Gather": 149, "Slice": 123, "LayerNormalization": 120, "Sigmoid": 96, "ConvInteger": 77, "Where": 73, "MatMul": 72, "ConstantOfShape": 51, "Pad": 48, "Div": 27, "Squeeze": 24, "Softmax": 24, "Split": 24, "Floor": 3, "Relu": 3, "Sub": 2, "Expand": 2, "Not": 2, "And": 2 }, "op_risk_report": { "rows": [ { "op": "ConstantOfShape", "count": 51, "risk": "memory_amp", "note": "Memory amplification — allocates tensor of declared shape" }, { "op": "Expand", "count": 2, "risk": "memory_amp", "note": "Memory amplification — broadcasts to declared shape" }, { "op": "Tile", "count": 1, "risk": "memory_amp", "note": "Memory amplification — replicates input N times" }, { "op": "Gather", "count": 149, "risk": "runtime_sensitive", "note": "Runtime-sensitive — index-based gather" }, { "op": "MatMulInteger", "count": 217, "risk": "quantized", "note": "Quantized model — integer matmul" }, { "op": "ConvInteger", "count": 77, "risk": "quantized", "note": "Quantized model — integer convolution" }, { "op": "Constant", "count": 1661, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Mul", "count": 733, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Cast", "count": 426, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Unsqueeze", "count": 328, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Reshape", "count": 295, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Transpose", "count": 244, "risk": "normal", "note": "Standard ONNX op" }, { "op": "DynamicQuantizeLinear", "count": 223, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Concat", "count": 221, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Add", "count": 205, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Shape", "count": 175, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Slice", "count": 123, "risk": "normal", "note": "Standard ONNX op" }, { "op": "LayerNormalization", "count": 120, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Sigmoid", "count": 96, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Where", "count": 73, "risk": "normal", "note": "Standard ONNX op" }, { "op": "MatMul", "count": 72, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Pad", "count": 48, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Div", "count": 27, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Squeeze", "count": 24, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Softmax", "count": 24, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Split", "count": 24, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Floor", "count": 3, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Relu", "count": 3, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Sub", "count": 2, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Not", "count": 2, "risk": "normal", "note": "Standard ONNX op" }, { "op": "And", "count": 2, "risk": "normal", "note": "Standard ONNX op" }, { "op": "Equal", "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" } ], "elevated_categories": [ "memory_amp", "runtime_sensitive", "quantized" ], "has_control_flow": false, "has_memory_amp": true, "has_provider_compat": false, "has_runtime_sensitive": true, "has_quantized": true, "has_nondeterministic": false, "has_string_ops": false }, "inputs": [ { "name": "audio_signal", "dtype": "FLOAT", "shape": [ "audio_signal_dynamic_axes_1", 128, "audio_signal_dynamic_axes_2" ] }, { "name": "length", "dtype": "INT64", "shape": [ "length_dynamic_axes_1" ] } ], "outputs": [ { "name": "outputs", "dtype": "FLOAT", "shape": [ "Transposeoutputs_dim_0", 1024, "Transposeoutputs_dim_2" ] }, { "name": "encoded_lengths", "dtype": "INT64", "shape": [ "length_dynamic_axes_1" ] } ], "shape_risk_report": { "inputs": [ { "name": "audio_signal", "dtype": "FLOAT", "shape": [ "audio_signal_dynamic_axes_1", 128, "audio_signal_dynamic_axes_2" ], "has_dynamic": true, "has_negative": false, "est_elements": 128 }, { "name": "length", "dtype": "INT64", "shape": [ "length_dynamic_axes_1" ], "has_dynamic": true, "has_negative": false, "est_elements": 1 } ], "outputs": [ { "name": "outputs", "dtype": "FLOAT", "shape": [ "Transposeoutputs_dim_0", 1024, "Transposeoutputs_dim_2" ], "has_dynamic": true, "has_negative": false, "est_elements": 1024 }, { "name": "encoded_lengths", "dtype": "INT64", "shape": [ "length_dynamic_axes_1" ], "has_dynamic": true, "has_negative": false, "est_elements": 1 } ], "any_dynamic_dims": true, "dynamic_dim_names": [ "Transposeoutputs_dim_0", "Transposeoutputs_dim_2", "audio_signal_dynamic_axes_1", "audio_signal_dynamic_axes_2", "length_dynamic_axes_1" ], "negative_dims_detected": false, "extreme_dims_detected": false, "input_elements_est": 129, "output_elements_est": 1025, "amplification_ratio": 7.95 }, "tensor_count": 1229, "param_estimate": 608855720, "weights_size_bytes": 609822622, "weights_size_mb": 581.57, "dtype_histogram": { "FLOAT": 322086, "UINT8": 608533542, "INT64": 92 }, "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": { "onnx.infer": "onnxruntime.quant" }, "inline_tensor_mismatches": 0, "zero_dim_initializers": 0, "extreme_dim_tensors": 0, "initializer_integrity": { "count": 1229, "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": 318, "total_float_tensors": 318, "values_scanned": 321792, "nan_count": 0, "inf_count": 0, "affected_tensors": [] } }, "declared_weight_bytes": 609822622, "declared_weight_mb": 581.57, "weight_delta_bytes": 42361377, "weight_delta_mb": 40.4, "file_size_bytes": 652183999, "file_size_mb": 621.97, "sha256": "6139d2fa7e1b086097b277c7149725edbab89cc7c7ae64b23c741be4055aff09", "sha512": "90d6b7163914efd49ad006071282404b788911b2b26f8a5fe70407f193bbce2194a90a92c9f63cec9b7c810146889deb443ffdaf9eb9b0a1cd5b592023ab6be5", "dtype_stats": { "FLOAT": { "count": 612, "bytes": 1288344, "mb": 1.23, "avg_bits_per_elem": 32.0 }, "UINT8": { "count": 588, "bytes": 608533542, "mb": 580.34, "avg_bits_per_elem": 8.0 }, "INT64": { "count": 29, "bytes": 736, "mb": 0.0, "avg_bits_per_elem": 64.0 } }, "tensor_preview": [ { "index": 0, "name": "pre_encode.out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 1, "name": "pre_encode.conv.0.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 2, "name": "pre_encode.conv.2.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 3, "name": "pre_encode.conv.3.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 4, "name": "pre_encode.conv.5.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 5, "name": "pre_encode.conv.6.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 6, "name": "layers.0.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 7, "name": "layers.0.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 8, "name": "layers.0.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 9, "name": "layers.0.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 10, "name": "layers.0.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 11, "name": "layers.0.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 12, "name": "layers.0.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 13, "name": "layers.0.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 14, "name": "layers.0.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 15, "name": "layers.0.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 16, "name": "layers.0.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 17, "name": "layers.0.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 18, "name": "layers.1.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 19, "name": "layers.1.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 20, "name": "layers.1.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 21, "name": "layers.1.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 22, "name": "layers.1.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 23, "name": "layers.1.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 24, "name": "layers.1.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 25, "name": "layers.1.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 26, "name": "layers.1.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 27, "name": "layers.1.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 28, "name": "layers.1.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 29, "name": "layers.1.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 30, "name": "layers.2.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 31, "name": "layers.2.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 32, "name": "layers.2.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 33, "name": "layers.2.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 34, "name": "layers.2.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 35, "name": "layers.2.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 36, "name": "layers.2.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 37, "name": "layers.2.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 38, "name": "layers.2.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 39, "name": "layers.2.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 40, "name": "layers.2.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 41, "name": "layers.2.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 42, "name": "layers.3.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 43, "name": "layers.3.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 44, "name": "layers.3.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 45, "name": "layers.3.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 46, "name": "layers.3.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 47, "name": "layers.3.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 48, "name": "layers.3.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 49, "name": "layers.3.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 } ], "op_histogram_full": { "Constant": 1661, "Mul": 733, "Cast": 426, "Unsqueeze": 328, "Reshape": 295, "Transpose": 244, "DynamicQuantizeLinear": 223, "Concat": 221, "MatMulInteger": 217, "Add": 205, "Shape": 175, "Gather": 149, "Slice": 123, "LayerNormalization": 120, "Sigmoid": 96, "ConvInteger": 77, "Where": 73, "MatMul": 72, "ConstantOfShape": 51, "Pad": 48, "Div": 27, "Squeeze": 24, "Softmax": 24, "Split": 24, "Floor": 3, "Relu": 3, "Sub": 2, "Expand": 2, "Not": 2, "And": 2, "Equal": 1, "Range": 1, "Less": 1, "Tile": 1 } }, "opsets": { "ai.onnx": 17 }, "producer": "onnx.quantize", "domain": "", "ir_version": 8, "nodes": 5654, "initialized_tensors": 1229, "weights_size_est_bytes": 609822622, "inputs": [ { "name": "audio_signal", "dtype": "FLOAT", "shape": [ "audio_signal_dynamic_axes_1", 128, "audio_signal_dynamic_axes_2" ] }, { "name": "length", "dtype": "INT64", "shape": [ "length_dynamic_axes_1" ] } ], "outputs": [ { "name": "outputs", "dtype": "FLOAT", "shape": [ "Transposeoutputs_dim_0", 1024, "Transposeoutputs_dim_2" ] }, { "name": "encoded_lengths", "dtype": "INT64", "shape": [ "length_dynamic_axes_1" ] } ], "op_hist_top": { "Constant": 1661, "Mul": 733, "Cast": 426, "Unsqueeze": 328, "Reshape": 295, "Transpose": 244, "DynamicQuantizeLinear": 223, "Concat": 221, "MatMulInteger": 217, "Add": 205, "Shape": 175, "Gather": 149, "Slice": 123, "LayerNormalization": 120, "Sigmoid": 96, "ConvInteger": 77, "Where": 73, "MatMul": 72, "ConstantOfShape": 51, "Pad": 48, "Div": 27, "Squeeze": 24, "Softmax": 24, "Split": 24, "Floor": 3, "Relu": 3, "Sub": 2, "Expand": 2, "Not": 2, "And": 2 }, "dtype_hist": { "FLOAT": 322086, "UINT8": 608533542, "INT64": 92 }, "custom_domains": [], "has_external_data": false, "external_refs": [], "quantized": false, "model_type": "audio-speech", "file_size_mb": 621.97, "declared_weight_bytes": 609822622, "declared_weight_mb": 581.57, "weight_delta_mb": 40.4, "dtype_stats": { "FLOAT": { "count": 612, "bytes": 1288344, "mb": 1.23, "avg_bits_per_elem": 32.0 }, "UINT8": { "count": 588, "bytes": 608533542, "mb": 580.34, "avg_bits_per_elem": 8.0 }, "INT64": { "count": 29, "bytes": 736, "mb": 0.0, "avg_bits_per_elem": 64.0 } }, "tensor_preview": [ { "index": 0, "name": "pre_encode.out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 1, "name": "pre_encode.conv.0.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 2, "name": "pre_encode.conv.2.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 3, "name": "pre_encode.conv.3.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 4, "name": "pre_encode.conv.5.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 5, "name": "pre_encode.conv.6.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 6, "name": "layers.0.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 7, "name": "layers.0.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 8, "name": "layers.0.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 9, "name": "layers.0.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 10, "name": "layers.0.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 11, "name": "layers.0.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 12, "name": "layers.0.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 13, "name": "layers.0.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 14, "name": "layers.0.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 15, "name": "layers.0.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 16, "name": "layers.0.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 17, "name": "layers.0.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 18, "name": "layers.1.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 19, "name": "layers.1.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 20, "name": "layers.1.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 21, "name": "layers.1.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 22, "name": "layers.1.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 23, "name": "layers.1.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 24, "name": "layers.1.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 25, "name": "layers.1.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 26, "name": "layers.1.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 27, "name": "layers.1.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 28, "name": "layers.1.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 29, "name": "layers.1.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 30, "name": "layers.2.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 31, "name": "layers.2.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 32, "name": "layers.2.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 33, "name": "layers.2.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 34, "name": "layers.2.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 35, "name": "layers.2.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 36, "name": "layers.2.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 37, "name": "layers.2.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 38, "name": "layers.2.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 39, "name": "layers.2.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 40, "name": "layers.2.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 41, "name": "layers.2.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 42, "name": "layers.3.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 43, "name": "layers.3.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 44, "name": "layers.3.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 45, "name": "layers.3.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 46, "name": "layers.3.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 47, "name": "layers.3.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 48, "name": "layers.3.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 49, "name": "layers.3.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 } ] }, "tokenizer": {}, "tensors": { "count": 1229, "preview": [ { "index": 0, "name": "pre_encode.out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 1, "name": "pre_encode.conv.0.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 2, "name": "pre_encode.conv.2.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 3, "name": "pre_encode.conv.3.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 4, "name": "pre_encode.conv.5.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 5, "name": "pre_encode.conv.6.bias", "dims": [ 256 ], "dtype_name": "FLOAT", "byte_len": 1024, "bits_per_elem": 32.0 }, { "index": 6, "name": "layers.0.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 7, "name": "layers.0.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 8, "name": "layers.0.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 9, "name": "layers.0.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 10, "name": "layers.0.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 11, "name": "layers.0.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 12, "name": "layers.0.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 13, "name": "layers.0.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 14, "name": "layers.0.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 15, "name": "layers.0.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 16, "name": "layers.0.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 17, "name": "layers.0.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 18, "name": "layers.1.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 19, "name": "layers.1.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 20, "name": "layers.1.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 21, "name": "layers.1.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 22, "name": "layers.1.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 23, "name": "layers.1.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 24, "name": "layers.1.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 25, "name": "layers.1.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 26, "name": "layers.1.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 27, "name": "layers.1.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 28, "name": "layers.1.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 29, "name": "layers.1.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 30, "name": "layers.2.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 31, "name": "layers.2.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 32, "name": "layers.2.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 33, "name": "layers.2.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 34, "name": "layers.2.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 35, "name": "layers.2.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 36, "name": "layers.2.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 37, "name": "layers.2.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 38, "name": "layers.2.norm_feed_forward2.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 39, "name": "layers.2.norm_feed_forward2.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 40, "name": "layers.2.norm_out.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 41, "name": "layers.2.norm_out.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 42, "name": "layers.3.norm_feed_forward1.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 43, "name": "layers.3.norm_feed_forward1.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 44, "name": "layers.3.norm_conv.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 45, "name": "layers.3.norm_conv.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 46, "name": "layers.3.norm_self_att.weight", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 47, "name": "layers.3.norm_self_att.bias", "dims": [ 1024 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 48, "name": "layers.3.self_attn.pos_bias_u", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 }, { "index": 49, "name": "layers.3.self_attn.pos_bias_v", "dims": [ 8, 128 ], "dtype_name": "FLOAT", "byte_len": 4096, "bits_per_elem": 32.0 } ], "dtype_histogram": { "FLOAT": 322086, "UINT8": 608533542, "INT64": 92 }, "dtype_stats": { "FLOAT": { "count": 612, "bytes": 1288344, "mb": 1.23, "avg_bits_per_elem": 32.0 }, "UINT8": { "count": 588, "bytes": 608533542, "mb": 580.34, "avg_bits_per_elem": 8.0 }, "INT64": { "count": 29, "bytes": 736, "mb": 0.0, "avg_bits_per_elem": 64.0 } }, "integrity": { "declared_weight_bytes": 609822622, "weights_size_est_bytes": 609822622, "weight_delta_mb": 40.4, "has_external_data": false, "external_refs": [] } }, "security": { "grade": "B", "score": 91, "findings": [ { "severity": "INFO", "category": "Shape Risk", "title": "Dynamic tensor dimensions detected", "detail": "Inputs with dynamic dims: ['audio_signal', 'length']. Outputs: ['outputs', 'encoded_lengths'] (symbolic: Transposeoutputs_dim_0, Transposeoutputs_dim_2, audio_signal_dynamic_axes_1, audio_signal_dynamic_axes_2). 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": "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": "Resources", "title": "Large ONNX graph — 5,654 graph nodes / ~609M parameters", "detail": "This model has 5,654 nodes, 1229 initializers, 608,855,720 parameters, and 609.8 MB of declared weights. Larger graphs have higher memory allocation overhead at load time and may hit per-request limits in serving frameworks.", "recommendation": "Validate memory and latency under representative maximum input length. Set request concurrency and input-size limits before exposing the model as a service." }, { "severity": "INFO", "category": "Provenance", "title": "Model producer: onnx.quantize 0.1.0", "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 (5654 total nodes, 34 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 (1 input(s))", "detail": "Inputs with all-dynamic dimensions: [('length', \"['length_dynamic_axes_1']\")]. 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 | 5654 nodes | 608,855,720 params | 3 warnings / 5 informational", "bullets": [ "ONNX audio-speech model | opset 17 | 5654 nodes | 608,855,720 params | 3 warnings / 5 informational", "Size: 621.97 MB", "SHA256: 6139d2fa7e1b0860..." ], "recommendations": [ "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." ] } }