neural_compressor.ux.utils.parser

Parsers for log files.

Module Contents

Classes

Parser	Parser abstract class.
OptimizationParser	Parser class is responsible for parsing optimization log files.
PerformanceParser	Parser class is responsible for parsing performance benchmark log files.
AccuracyParser	Parser class is responsible for parsing accuracy benchmark log files.
BenchmarkParserFactory	Benchmark parser factory.
ProfilingParser	Parser class is responsible for parsing profiling log files.

class neural_compressor.ux.utils.parser.Parser(logs: list)

Bases: abc.ABC

Parser abstract class.

abstract property patterns: dict

Set patterns to get metrics from lines.

abstract process() → Dict[str, Any]

Process log files.

class neural_compressor.ux.utils.parser.OptimizationParser(logs: list)

Bases: Parser

Parser class is responsible for parsing optimization log files.

property patterns: dict

Set patterns to get metrics from lines.

process() → Dict[str, Any]

Process files.

class neural_compressor.ux.utils.parser.PerformanceParser(logs: list)

Bases: Parser

Parser class is responsible for parsing performance benchmark log files.

property patterns: dict

Set patterns to get metrics from lines.

process() → Dict[str, Any]

Process files.

static update_partial(partial: Dict[str, List], parsed_result: Dict[str, float | int]) → Dict[str, List]

Update partial entries.

summarize_partial(partial: dict) → dict

Calculate final values.

static summarize_value(key: str, value: list) → float | int

Calculate final value.

class neural_compressor.ux.utils.parser.AccuracyParser(logs: list)

Bases: Parser

Parser class is responsible for parsing accuracy benchmark log files.

property patterns: dict

Set patterns to get metrics from lines.

process() → Dict[str, Any]

Process accuracy logs.

class neural_compressor.ux.utils.parser.BenchmarkParserFactory

Benchmark parser factory.

static get_parser(benchmark_mode: str, logs: List[str]) → Parser

Get benchmark parser for specified mode.

class neural_compressor.ux.utils.parser.ProfilingParser(logs: list)

Bases: Parser

Parser class is responsible for parsing profiling log files.

property patterns: dict

Set patterns to get metrics from lines.

static unify_time(string_value: str) → float

Unify time with unit to micro seconds float value.

Parameters:

string_value (str) – time value with unit, e.g. 125.6ms

Returns:

time value in microseconds

Return type:

float

process() → Dict[str, Any]

Process profiling logs.