Skip to content

CLI

cli

Command-line interface for pytorch-ir.

create_parser 

create_parser() -> argparse.ArgumentParser

Create the argument parser for pytorch-ir CLI.

Source code in torch_ir/cli.py 
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
def create_parser() -> argparse.ArgumentParser:
    """Create the argument parser for pytorch-ir CLI."""
    parser = argparse.ArgumentParser(prog="pytorch-ir", description="pytorch-ir CLI tools")
    parser.add_argument("--version", action="version", version=f"pytorch-ir {__version__}")

    subparsers = parser.add_subparsers(dest="command")

    # visualize subcommand
    vis_parser = subparsers.add_parser("visualize", help="Visualize IR graph as Mermaid diagram")
    vis_parser.add_argument("ir_file", help="Path to IR JSON file")
    vis_parser.add_argument(
        "--max-nodes",
        type=int,
        default=30,
        help="Maximum number of nodes to display (default: 30)",
    )
    vis_parser.add_argument(
        "--no-weights",
        action="store_true",
        help="Hide weight inputs from the diagram",
    )
    vis_parser.add_argument(
        "-o",
        "--output",
        default=None,
        help="Output file (.png/.svg for image, others for Mermaid text)",
    )

    # info subcommand
    info_parser = subparsers.add_parser("info", help="Show IR summary information")
    info_parser.add_argument("ir_file", help="Path to IR JSON file")
    info_parser.add_argument("--json", action="store_true", help="Output in JSON format")
    info_parser.add_argument("-o", "--output", default=None, help="Output file")

    return parser

main 

main(argv: Optional[List[str]] = None) -> None

CLI entry point.

Source code in torch_ir/cli.py 
174
175
176
177
178
179
180
181
182
183
184
185
186
def main(argv: Optional[List[str]] = None) -> None:
    """CLI entry point."""
    parser = create_parser()
    args = parser.parse_args(argv)

    if args.command is None:
        parser.print_help()
        sys.exit(2)

    if args.command == "visualize":
        cmd_visualize(args)
    elif args.command == "info":
        cmd_info(args)

cmd_visualize 

cmd_visualize(args: Namespace) -> None

Handle the 'visualize' subcommand.

Source code in torch_ir/cli.py 
66
67
68
69
70
71
72
73
74
75
76
77
def cmd_visualize(args: argparse.Namespace) -> None:
    """Handle the 'visualize' subcommand."""
    ir = _load_ir_file(args.ir_file)
    mermaid_text = ir_to_mermaid(ir, max_nodes=args.max_nodes, include_weights=not args.no_weights)

    if args.output is not None:
        ext = Path(args.output).suffix.lower()
        if ext in (".png", ".svg"):
            _render_mermaid_image(mermaid_text, args.output)
            return

    _write_output(mermaid_text, args.output)

cmd_info 

cmd_info(args: Namespace) -> None

Handle the 'info' subcommand.

Source code in torch_ir/cli.py 
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
def cmd_info(args: argparse.Namespace) -> None:
    """Handle the 'info' subcommand."""
    ir = _load_ir_file(args.ir_file)

    # Compute op distribution
    op_counts: dict[str, int] = defaultdict(int)
    for node in ir.nodes:
        op_counts[node.op_type] += 1

    # Compute total weight parameters
    total_params = 0
    for w in ir.weights:
        p = 1
        for d in w.shape:
            p *= d
        total_params += p

    info = {
        "model_name": ir.model_name,
        "num_nodes": len(ir.nodes),
        "num_inputs": len(ir.graph_inputs),
        "num_outputs": len(ir.graph_outputs),
        "num_weights": len(ir.weights),
        "total_parameters": total_params,
        "input_shapes": {inp.name: list(inp.shape) for inp in ir.graph_inputs},
        "output_shapes": {out.name: list(out.shape) for out in ir.graph_outputs},
        "op_distribution": dict(sorted(op_counts.items(), key=lambda x: -x[1])),
    }

    if args.json:
        output = json.dumps(info, indent=2)
    else:
        lines = [
            f"Model: {info['model_name']}",
            f"Nodes: {info['num_nodes']}",
            f"Inputs: {info['num_inputs']}",
            f"Outputs: {info['num_outputs']}",
            f"Weights: {info['num_weights']}",
            f"Total parameters: {info['total_parameters']:,}",
            "",
            "Input shapes:",
        ]
        for name, shape in info["input_shapes"].items():
            lines.append(f"  {name}: {shape}")
        lines.append("")
        lines.append("Output shapes:")
        for name, shape in info["output_shapes"].items():
            lines.append(f"  {name}: {shape}")
        lines.append("")
        lines.append("Op distribution:")
        for op, count in info["op_distribution"].items():
            lines.append(f"  {op}: {count}")
        output = "\n".join(lines)

    _write_output(output, args.output)