haoline 0.3.0__py3-none-any.whl

haoline/.streamlit/config.toml

@@ -0,0 +1,10 @@
+[theme]
+base = "dark"
+primaryColor = "#10b981"
+backgroundColor = "#0d0d0d"
+secondaryBackgroundColor = "#161616"
+textColor = "#f5f5f5"
+
+[server]
+maxUploadSize = 500
+

haoline/__init__.py

@@ -0,0 +1,248 @@
+# Copyright (c) 2025 HaoLine Contributors
+# SPDX-License-Identifier: MIT
+
+"""
+HaoLine (皓线) - Universal Model Inspector.
+
+See what's really inside your models.
+
+This module provides utilities for analyzing neural network models to extract:
+- Parameter counts and memory estimates
+- FLOP estimates per operation
+- Architectural pattern detection (transformers, CNNs, residual blocks)
+- Risk signals for deployment
+- Hardware performance estimates
+
+Supports: ONNX, PyTorch, TensorFlow, TensorRT
+
+Example usage:
+    from haoline import ModelInspector
+
+    inspector = ModelInspector()
+    report = inspector.inspect("model.onnx")
+    print(report.to_json())
+"""
+
+__version__ = "0.3.0"
+
+from .analyzer import MetricsEngine, ONNXGraphLoader
+from .compare_visualizations import (
+    CalibrationRecommendation,
+    LayerPrecisionBreakdown,
+    NormalizedMetrics,
+    TradeoffPoint,
+    analyze_tradeoffs,
+    build_enhanced_markdown,
+    compute_normalized_metrics,
+    compute_tradeoff_points,
+    extract_layer_precision_breakdown,
+    generate_calibration_recommendations,
+    generate_compare_html,
+    generate_compare_pdf,
+    generate_layer_precision_chart,
+    generate_memory_savings_chart,
+    generate_radar_chart,
+    generate_tradeoff_chart,
+)
+from .compare_visualizations import is_available as is_compare_viz_available
+from .edge_analysis import EdgeAnalysisResult, EdgeAnalyzer
+from .hardware import (
+    HARDWARE_PROFILES,
+    HardwareDetector,
+    HardwareEstimates,
+    HardwareEstimator,
+    HardwareProfile,
+    detect_local_hardware,
+    get_profile,
+    list_available_profiles,
+)
+from .hierarchical_graph import (
+    HierarchicalGraph,
+    HierarchicalGraphBuilder,
+    HierarchicalNode,
+)
+from .hierarchical_graph import generate_summary as generate_graph_summary
+from .html_export import HTMLExporter
+from .html_export import generate_html as generate_graph_html
+from .layer_summary import (
+    LayerMetrics,
+    LayerSummary,
+    LayerSummaryBuilder,
+    generate_html_table,
+    generate_markdown_table,
+)
+from .llm_summarizer import LLMSummarizer, LLMSummary, summarize_report
+from .llm_summarizer import has_api_key as has_llm_api_key
+from .llm_summarizer import is_available as is_llm_available
+from .operational_profiling import (
+    BatchSizeSweep,
+    BatchSweepPoint,
+    BottleneckAnalysis,
+    GPUMetrics,
+    LayerProfile,
+    OperationalProfiler,
+    ProfilingResult,
+    ResolutionPoint,
+    ResolutionSweep,
+    SystemRequirements,
+)
+from .patterns import PatternAnalyzer
+from .pdf_generator import PDFGenerator, generate_pdf
+from .pdf_generator import is_available as is_pdf_available
+from .report import (
+    DatasetInfo,
+    InspectionReport,
+    ModelInspector,
+    infer_num_classes_from_output,
+)
+from .report_sections import (
+    BlockSummaryItem,
+    BottleneckSection,
+    DetectedBlocksSection,
+    ExtractedReportSections,
+    HardwareEstimatesSection,
+    KVCacheSection,
+    MemoryBreakdownRow,
+    MemoryBreakdownSection,
+    MetricsCard,
+    MetricsSummary,
+    OperatorDistribution,
+    PrecisionBreakdown,
+    PrecisionBreakdownRow,
+    RiskSignalItem,
+    RiskSignalsSection,
+    SharedWeightsSection,
+    format_bytes,
+    format_number,
+)
+from .risks import RiskAnalyzer, RiskSignal, RiskThresholds
+from .schema import (
+    INSPECTION_REPORT_SCHEMA,
+    ValidationError,
+    get_schema,
+    validate_report,
+    validate_report_strict,
+)
+from .universal_ir import (
+    DataType,
+    GraphMetadata,
+    SourceFormat,
+    TensorOrigin,
+    UniversalGraph,
+    UniversalNode,
+    UniversalTensor,
+)
+from .visualizations import (
+    THEME,
+    ChartTheme,
+    VisualizationGenerator,
+    generate_visualizations,
+)
+from .visualizations import is_available as is_visualization_available
+
+__all__ = [
+    # Universal IR (Epic 18)
+    "DataType",
+    "GraphMetadata",
+    "SourceFormat",
+    "TensorOrigin",
+    "UniversalGraph",
+    "UniversalNode",
+    "UniversalTensor",
+    # Core
+    "HARDWARE_PROFILES",
+    "INSPECTION_REPORT_SCHEMA",
+    "THEME",
+    "BatchSizeSweep",
+    "BatchSweepPoint",
+    "BottleneckAnalysis",
+    "CalibrationRecommendation",
+    "ChartTheme",
+    "DatasetInfo",
+    "EdgeAnalysisResult",
+    "EdgeAnalyzer",
+    "GPUMetrics",
+    "HTMLExporter",
+    "HardwareDetector",
+    "HardwareEstimates",
+    "HardwareEstimator",
+    "HardwareProfile",
+    "HierarchicalGraph",
+    "HierarchicalGraphBuilder",
+    "HierarchicalNode",
+    "InspectionReport",
+    "LLMSummarizer",
+    "LLMSummary",
+    "LayerMetrics",
+    "LayerPrecisionBreakdown",
+    "LayerProfile",
+    "LayerSummary",
+    "LayerSummaryBuilder",
+    "MetricsEngine",
+    "ModelInspector",
+    "NormalizedMetrics",
+    "ONNXGraphLoader",
+    "OperationalProfiler",
+    "PDFGenerator",
+    "PatternAnalyzer",
+    "ProfilingResult",
+    "ResolutionPoint",
+    "ResolutionSweep",
+    "RiskAnalyzer",
+    "RiskSignal",
+    "RiskThresholds",
+    "SystemRequirements",
+    "TradeoffPoint",
+    "ValidationError",
+    "VisualizationGenerator",
+    "analyze_tradeoffs",
+    "build_enhanced_markdown",
+    "compute_normalized_metrics",
+    "compute_tradeoff_points",
+    "detect_local_hardware",
+    "extract_layer_precision_breakdown",
+    "generate_calibration_recommendations",
+    "generate_compare_html",
+    "generate_compare_pdf",
+    "generate_graph_html",
+    "generate_graph_summary",
+    "generate_html_table",
+    "generate_layer_precision_chart",
+    "generate_markdown_table",
+    "generate_memory_savings_chart",
+    "generate_pdf",
+    "generate_radar_chart",
+    "generate_tradeoff_chart",
+    "generate_visualizations",
+    "get_profile",
+    "get_schema",
+    "has_llm_api_key",
+    "infer_num_classes_from_output",
+    "is_compare_viz_available",
+    "is_llm_available",
+    "is_pdf_available",
+    "is_visualization_available",
+    "list_available_profiles",
+    "summarize_report",
+    "validate_report",
+    "validate_report_strict",
+    # Report Sections (Story 41.2)
+    "BlockSummaryItem",
+    "BottleneckSection",
+    "DetectedBlocksSection",
+    "ExtractedReportSections",
+    "HardwareEstimatesSection",
+    "KVCacheSection",
+    "MemoryBreakdownRow",
+    "MemoryBreakdownSection",
+    "MetricsCard",
+    "MetricsSummary",
+    "OperatorDistribution",
+    "PrecisionBreakdown",
+    "PrecisionBreakdownRow",
+    "RiskSignalItem",
+    "RiskSignalsSection",
+    "SharedWeightsSection",
+    "format_bytes",
+    "format_number",
+]