quanto

1
# Copyright 2024 The HuggingFace Team. All rights reserved.
2
#
3
# Licensed under the Apache License, Version 2.0 (the "License");
4
# you may not use this file except in compliance with the License.
5
# You may obtain a copy of the License at
6
#
7
#     http://www.apache.org/licenses/LICENSE-2.0
8
#
9
# Unless required by applicable law or agreed to in writing, software
10
# distributed under the License is distributed on an "AS IS" BASIS,
11
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
# See the License for the specific language governing permissions and
13
# limitations under the License.
14

15
from contextlib import nullcontext
16

17
import pytest
18
import torch
19

20
from quanto.library import disable_extensions
21
from quanto.tensor.packed import pack_weights
22

23

24
@pytest.mark.parametrize("bits", [2, 4], ids=["int2", "int4"])
25
@pytest.mark.parametrize("shape", [(12,), (32, 32)], ids=["vector", "matrix"])
26
@pytest.mark.parametrize("use_ext", [True, False], ids=["ext", "no-ext"])
27
def test_unpack(bits, shape, use_ext, device):
28
    qmax = 2**bits
29
    a = torch.randint(0, qmax, shape, dtype=torch.uint8).to(device)
30
    packed_a = pack_weights(a, bits)
31
    context = nullcontext() if use_ext else disable_extensions()
32
    with context:
33
        unpacked_a = torch.ops.quanto.unpack(packed_a, bits)
34
    assert unpacked_a.dtype == torch.uint8
35
    assert torch.equal(unpacked_a, a)
36