lightseekorg--tokenspeed/python/tokenspeed/runtime/layers/quantization/__init__.py

# Copyright (c) 2026 LightSeek Foundation
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.

from enum import Enum

from tokenspeed.runtime.layers.quantization.base_config import (  # noqa: F401
    QuantizationConfig,
    QuantizeMethodBase,
)
from tokenspeed.runtime.layers.quantization.compressed_tensors.compressed_tensors import (
    CompressedTensorsConfig,
)
from tokenspeed.runtime.layers.quantization.fp8 import Fp8Config
from tokenspeed.runtime.layers.quantization.mxfp4 import Mxfp4Config
from tokenspeed.runtime.layers.quantization.nvfp4 import Nvfp4Config
from tokenspeed.runtime.layers.quantization.w8a8_fp8 import W8A8Fp8Config

BASE_QUANTIZATION_METHODS: dict[str, type[QuantizationConfig]] = {
    "fp8": Fp8Config,
    "w8a8_fp8": W8A8Fp8Config,
    "compressed-tensors": CompressedTensorsConfig,
    "nvfp4": Nvfp4Config,
    "mxfp4": Mxfp4Config,
}


QUANTIZATION_METHODS = BASE_QUANTIZATION_METHODS


def get_quantization_config(quantization: str) -> type[QuantizationConfig]:
    if quantization not in QUANTIZATION_METHODS:
        raise ValueError(
            f"Invalid quantization method: {quantization}. "
            f"Available methods: {list(QUANTIZATION_METHODS.keys())}"
        )
    return QUANTIZATION_METHODS[quantization]


class FusedMoeWeightScaleSupported(Enum):
    TENSOR = "tensor"
    CHANNEL = "channel"
    GROUP = "group"
    BLOCK = "block"