diff --git a/CHANGELOG.md b/CHANGELOG.md
index 69f32f1c..83ebbf97 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,14 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+
+## Unreleased
+
+### Added
+- `use_gpu` for PureSVD ([#229](https://github.com/MobileTeleSystems/RecTools/pull/229))
+- `from_params` method for models and `model_from_params` function ([#252](https://github.com/MobileTeleSystems/RecTools/pull/252))
+
+
 ## [0.10.0] - 16.01.2025
 
 ### Added
diff --git a/poetry.lock b/poetry.lock
index d73fdc78..93d95540 100644
--- a/poetry.lock
+++ b/poetry.lock
@@ -529,6 +529,36 @@ tomli = {version = "*", optional = true, markers = "python_full_version <= \"3.1
 [package.extras]
 toml = ["tomli"]
 
+[[package]]
+name = "cupy-cuda12x"
+version = "13.3.0"
+description = "CuPy: NumPy & SciPy for GPU"
+optional = true
+python-versions = ">=3.9"
+files = [
+    {file = "cupy_cuda12x-13.3.0-cp310-cp310-manylinux2014_aarch64.whl", hash = "sha256:674488e990998042cc54d2486d3c37cae80a12ba3787636be5a10b9446dd6914"},
+    {file = "cupy_cuda12x-13.3.0-cp310-cp310-manylinux2014_x86_64.whl", hash = "sha256:cf4a2a0864364715881b50012927e88bd7ec1e6f1de3987970870861ae5ed25e"},
+    {file = "cupy_cuda12x-13.3.0-cp310-cp310-win_amd64.whl", hash = "sha256:7c0dc8c49d271d1c03e49a5d6c8e42e8fee3114b10f269a5ecc387731d693eaa"},
+    {file = "cupy_cuda12x-13.3.0-cp311-cp311-manylinux2014_aarch64.whl", hash = "sha256:c0cc095b9a3835fd5db66c45ed3c58ecdc5a3bb14e53e1defbfd4a0ce5c8ecdb"},
+    {file = "cupy_cuda12x-13.3.0-cp311-cp311-manylinux2014_x86_64.whl", hash = "sha256:a0e3bead04e502ebde515f0343444ca3f4f7aed09cbc3a316a946cba97f2ea66"},
+    {file = "cupy_cuda12x-13.3.0-cp311-cp311-win_amd64.whl", hash = "sha256:5f11df1149c7219858b27e4c8be92cb4eaf7364c94af6b78c40dffb98050a61f"},
+    {file = "cupy_cuda12x-13.3.0-cp312-cp312-manylinux2014_aarch64.whl", hash = "sha256:bbd0d916310391faf0d7dc9c58fff7a6dc996b67e5768199160bbceb5ebdda8c"},
+    {file = "cupy_cuda12x-13.3.0-cp312-cp312-manylinux2014_x86_64.whl", hash = "sha256:e206bd8664f0763732b6012431f484ee535bffd77a5ae95e9bfe1c7c72396625"},
+    {file = "cupy_cuda12x-13.3.0-cp312-cp312-win_amd64.whl", hash = "sha256:88ef1478f00ae252da0026e7f04f70c9bb6a2dc130ba5f1e5bc5e8069a928bf5"},
+    {file = "cupy_cuda12x-13.3.0-cp39-cp39-manylinux2014_aarch64.whl", hash = "sha256:3a52aa49ffcc940d034f2bb39728c90e9fa83c7a49e376404507956adb6d6ec4"},
+    {file = "cupy_cuda12x-13.3.0-cp39-cp39-manylinux2014_x86_64.whl", hash = "sha256:3ef13f3cbc449d2a0f816594ab1fa0236e1f06ad1eaa81ad04c75e47cbeb87be"},
+    {file = "cupy_cuda12x-13.3.0-cp39-cp39-win_amd64.whl", hash = "sha256:8f5433eec3e5cd8d39e8fcb82e0fdab7c22eba8e3304fcb0b42f2ea988fef0d6"},
+]
+
+[package.dependencies]
+fastrlock = ">=0.5"
+numpy = ">=1.22,<2.3"
+
+[package.extras]
+all = ["Cython (>=0.29.22,<3)", "optuna (>=2.0)", "scipy (>=1.7,<1.14)"]
+stylecheck = ["autopep8 (==1.5.5)", "flake8 (==3.8.4)", "mypy (==1.4.1)", "pbr (==5.5.1)", "pycodestyle (==2.6.0)", "types-setuptools (==57.4.14)"]
+test = ["hypothesis (>=6.37.2,<6.55.0)", "mpmath", "packaging", "pytest (>=7.2)"]
+
 [[package]]
 name = "decorator"
 version = "5.1.1"
@@ -597,6 +627,90 @@ files = [
 [package.extras]
 devel = ["colorama", "json-spec", "jsonschema", "pylint", "pytest", "pytest-benchmark", "pytest-cache", "validictory"]
 
+[[package]]
+name = "fastrlock"
+version = "0.8.2"
+description = "Fast, re-entrant optimistic lock implemented in Cython"
+optional = true
+python-versions = "*"
+files = [
+    {file = "fastrlock-0.8.2-cp27-cp27m-macosx_10_15_x86_64.whl", hash = "sha256:94e348c72a1fd1f8191f25ea056448e4f5a87b8fbf005b39d290dcb0581a48cd"},
+    {file = "fastrlock-0.8.2-cp27-cp27m-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:2d5595903444c854b99c42122b87edfe8a37cd698a4eae32f4fd1d2a7b6c115d"},
+    {file = "fastrlock-0.8.2-cp27-cp27m-manylinux_2_5_x86_64.manylinux1_x86_64.whl", hash = "sha256:e4bbde174a0aff5f6eeba75cf8c4c5d2a316316bc21f03a0bddca0fc3659a6f3"},
+    {file = "fastrlock-0.8.2-cp27-cp27mu-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:7a2ccaf88ac0db153e84305d1ef0aa138cea82c6a88309066f6eaa3bc98636cd"},
+    {file = "fastrlock-0.8.2-cp27-cp27mu-manylinux_2_5_x86_64.manylinux1_x86_64.whl", hash = "sha256:31a27a2edf482df72b91fe6c6438314d2c65290aa7becc55589d156c9b91f0da"},
+    {file = "fastrlock-0.8.2-cp310-cp310-macosx_11_0_x86_64.whl", hash = "sha256:e9904b5b37c3e5bb4a245c56bc4b7e497da57ffb8528f4fc39af9dcb168ee2e1"},
+    {file = "fastrlock-0.8.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_24_aarch64.whl", hash = "sha256:43a241655e83e4603a152192cf022d5ca348c2f4e56dfb02e5c9c4c1a32f9cdb"},
+    {file = "fastrlock-0.8.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:9121a894d74e65557e47e777060a495ab85f4b903e80dd73a3c940ba042920d7"},
+    {file = "fastrlock-0.8.2-cp310-cp310-manylinux_2_5_i686.manylinux1_i686.manylinux_2_24_i686.whl", hash = "sha256:11bbbbc526363955aeddb9eec4cee2a0012322b7b2f15b54f44454fcf4fd398a"},
+    {file = "fastrlock-0.8.2-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_24_x86_64.whl", hash = "sha256:27786c62a400e282756ae1b090bcd7cfa35f28270cff65a9e7b27a5327a32561"},
+    {file = "fastrlock-0.8.2-cp310-cp310-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:08315bde19d0c2e6b06593d5a418be3dc8f9b1ee721afa96867b9853fceb45cf"},
+    {file = "fastrlock-0.8.2-cp310-cp310-musllinux_1_1_aarch64.whl", hash = "sha256:e8b49b5743ede51e0bcf6805741f39f5e0e0fd6a172ba460cb39e3097ba803bb"},
+    {file = "fastrlock-0.8.2-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:b443e73a4dfc7b6e0800ea4c13567b9694358e86f53bb2612a51c9e727cac67b"},
+    {file = "fastrlock-0.8.2-cp310-cp310-win_amd64.whl", hash = "sha256:b3853ed4ce522598dc886160a7bab432a093051af85891fa2f5577c1dcac8ed6"},
+    {file = "fastrlock-0.8.2-cp311-cp311-macosx_10_15_universal2.whl", hash = "sha256:790fc19bccbd39426060047e53629f171a44745613bf360a045e9f9c8c4a2cea"},
+    {file = "fastrlock-0.8.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_24_aarch64.whl", hash = "sha256:dbdce852e6bb66e1b8c36679d482971d69d93acf1785657522e51b7de30c3356"},
+    {file = "fastrlock-0.8.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d47713ffe6d4a627fbf078be9836a95ac106b4a0543e3841572c91e292a5d885"},
+    {file = "fastrlock-0.8.2-cp311-cp311-manylinux_2_5_i686.manylinux1_i686.manylinux_2_24_i686.whl", hash = "sha256:ea96503b918fceaf40443182742b8964d47b65c5ebdea532893cb9479620000c"},
+    {file = "fastrlock-0.8.2-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_24_x86_64.whl", hash = "sha256:c6bffa978793bea5e1b00e677062e53a62255439339591b70e209fa1552d5ee0"},
+    {file = "fastrlock-0.8.2-cp311-cp311-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:75c07726c8b1a52147fd7987d6baaa318c5dced1416c3f25593e40f56e10755b"},
+    {file = "fastrlock-0.8.2-cp311-cp311-musllinux_1_1_aarch64.whl", hash = "sha256:88f079335e9da631efa64486c8207564a7bcd0c00526bb9e842e9d5b7e50a6cc"},
+    {file = "fastrlock-0.8.2-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:4fb2e77ff04bc4beb71d63c8e064f052ce5a6ea1e001d528d4d7f4b37d736f2e"},
+    {file = "fastrlock-0.8.2-cp311-cp311-win_amd64.whl", hash = "sha256:b4c9083ea89ab236b06e9ef2263971db3b4b507195fc7d5eecab95828dcae325"},
+    {file = "fastrlock-0.8.2-cp312-cp312-macosx_10_15_universal2.whl", hash = "sha256:98195866d3a9949915935d40a88e4f1c166e82e378f622c88025f2938624a90a"},
+    {file = "fastrlock-0.8.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b22ea9bf5f9fad2b0077e944a7813f91593a4f61adf8faf734a70aed3f2b3a40"},
+    {file = "fastrlock-0.8.2-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:dcc1bf0ac8a194313cf6e645e300a8a379674ceed8e0b1e910a2de3e3c28989e"},
+    {file = "fastrlock-0.8.2-cp312-cp312-musllinux_1_1_aarch64.whl", hash = "sha256:a3dcc876050b8f5cbc0ee84ef1e7f0c1dfe7c148f10098828bc4403683c33f10"},
+    {file = "fastrlock-0.8.2-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:685e656048b59d8dfde8c601f188ad53a4d719eb97080cafc8696cda6d75865e"},
+    {file = "fastrlock-0.8.2-cp312-cp312-win_amd64.whl", hash = "sha256:fb5363cf0fddd9b50525ddbf64a1e1b28ec4c6dfb28670a940cb1cf988a6786b"},
+    {file = "fastrlock-0.8.2-cp35-cp35m-macosx_10_15_x86_64.whl", hash = "sha256:a74f5a92fa6e51c4f3c69b29c4662088b97be12f40652a21109605a175c81824"},
+    {file = "fastrlock-0.8.2-cp35-cp35m-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:ccf39ad5702e33e4d335b48ef9d56e21619b529b7f7471b5211419f380329b62"},
+    {file = "fastrlock-0.8.2-cp35-cp35m-manylinux_2_5_x86_64.manylinux1_x86_64.whl", hash = "sha256:66f2662c640bb71a1016a031eea6eef9d25c2bcdf7ffd1d1ddc5a58f9a1ced04"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-macosx_10_15_x86_64.whl", hash = "sha256:17734e2e5af4c07ddb0fb10bd484e062c22de3be6b67940b9cc6ec2f18fa61ba"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_24_aarch64.whl", hash = "sha256:ab91b0c36e95d42e1041a4907e3eefd06c482d53af3c7a77be7e214cc7cd4a63"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b32fdf874868326351a75b1e4c02f97e802147119ae44c52d3d9da193ec34f5b"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-manylinux_2_5_i686.manylinux1_i686.manylinux_2_24_i686.whl", hash = "sha256:2074548a335fcf7d19ebb18d9208da9e33b06f745754466a7e001d2b1c58dd19"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:4fb04442b6d1e2b36c774919c6bcbe3339c61b337261d4bd57e27932589095af"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_24_x86_64.whl", hash = "sha256:1fed2f4797ad68e9982038423018cf08bec5f4ce9fed63a94a790773ed6a795c"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:e380ec4e6d8b26e389713995a43cb7fe56baea2d25fe073d4998c4821a026211"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-manylinux_2_5_x86_64.manylinux1_x86_64.whl", hash = "sha256:25945f962c7bd808415cfde3da624d4399d4ea71ed8918538375f16bceb79e1c"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-musllinux_1_1_aarch64.whl", hash = "sha256:2c1719ddc8218b01e82fb2e82e8451bd65076cb96d7bef4477194bbb4305a968"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-musllinux_1_1_x86_64.whl", hash = "sha256:5460c5ee6ced6d61ec8cd2324ebbe793a4960c4ffa2131ffff480e3b61c99ec5"},
+    {file = "fastrlock-0.8.2-cp36-cp36m-win_amd64.whl", hash = "sha256:33145acbad8317584cd64588131c7e1e286beef6280c0009b4544c91fce171d2"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_24_aarch64.whl", hash = "sha256:59344c1d46b7dec97d3f22f1cc930fafe8980b3c5bc9c9765c56738a5f1559e4"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:b2a1c354f13f22b737621d914f3b4a8434ae69d3027a775e94b3e671756112f9"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-manylinux_2_5_i686.manylinux1_i686.manylinux_2_24_i686.whl", hash = "sha256:cf81e0278b645004388873e0a1f9e3bc4c9ab8c18e377b14ed1a544be4b18c9a"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:1b15430b93d7eb3d56f6ff690d2ebecb79ed0e58248427717eba150a508d1cd7"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_24_x86_64.whl", hash = "sha256:067edb0a0805bf61e17a251d5046af59f6e9d2b8ad01222e0ef7a0b7937d5548"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:eb31fe390f03f7ae886dcc374f1099ec88526631a4cb891d399b68181f154ff0"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.whl", hash = "sha256:643e1e65b4f5b284427e61a894d876d10459820e93aa1e724dfb415117be24e0"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-musllinux_1_1_aarch64.whl", hash = "sha256:5dfb78dd600a12f23fc0c3ec58f81336229fdc74501ecf378d1ce5b3f2f313ea"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:b8ca0fe21458457077e4cb2d81e1ebdb146a00b3e9e2db6180a773f7ea905032"},
+    {file = "fastrlock-0.8.2-cp37-cp37m-win_amd64.whl", hash = "sha256:d918dfe473291e8bfd8e13223ea5cb9b317bd9f50c280923776c377f7c64b428"},
+    {file = "fastrlock-0.8.2-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_24_aarch64.whl", hash = "sha256:c393af77c659a38bffbca215c0bcc8629ba4299568308dd7e4ff65d62cabed39"},
+    {file = "fastrlock-0.8.2-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:73426f5eb2ecc10626c67cf86bd0af9e00d53e80e5c67d5ce8e18376d6abfa09"},
+    {file = "fastrlock-0.8.2-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.manylinux_2_24_i686.whl", hash = "sha256:320fd55bafee3eb069cfb5d6491f811a912758387ef2193840e2663e80e16f48"},
+    {file = "fastrlock-0.8.2-cp38-cp38-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:8c1c91a68926421f5ccbc82c85f83bd3ba593b121a46a1b9a554b3f0dd67a4bf"},
+    {file = "fastrlock-0.8.2-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_24_x86_64.whl", hash = "sha256:ad1bc61c7f6b0e58106aaab034916b6cb041757f708b07fbcdd9d6e1ac629225"},
+    {file = "fastrlock-0.8.2-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:87f4e01b042c84e6090dbc4fbe3415ddd69f6bc0130382323f9d3f1b8dd71b46"},
+    {file = "fastrlock-0.8.2-cp38-cp38-manylinux_2_5_x86_64.manylinux1_x86_64.whl", hash = "sha256:d34546ad2e4a480b94b6797bcc5a322b3c705c4c74c3e4e545c4a3841c1b2d59"},
+    {file = "fastrlock-0.8.2-cp38-cp38-musllinux_1_1_aarch64.whl", hash = "sha256:ebb32d776b61acd49f859a1d16b9e3d84e7b46d0d92aebd58acd54dc38e96664"},
+    {file = "fastrlock-0.8.2-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:30bdbe4662992348132d03996700e1cf910d141d629179b967b146a22942264e"},
+    {file = "fastrlock-0.8.2-cp38-cp38-win_amd64.whl", hash = "sha256:07ed3c7b3867c05a3d6be4ced200c7767000f3431b9be6da66972822dd86e8be"},
+    {file = "fastrlock-0.8.2-cp39-cp39-macosx_11_0_x86_64.whl", hash = "sha256:ddf5d247f686aec853ddcc9a1234bfcc6f57b0a0670d2ad82fc25d8ae7e6a15f"},
+    {file = "fastrlock-0.8.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_24_aarch64.whl", hash = "sha256:7269bb3fc15587b0c191eecd95831d771a7d80f0c48929e560806b038ff3066c"},
+    {file = "fastrlock-0.8.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:adcb9e77aa132cc6c9de2ffe7cf880a20aa8cdba21d367d1da1a412f57bddd5d"},
+    {file = "fastrlock-0.8.2-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.manylinux_2_24_i686.whl", hash = "sha256:a3b8b5d2935403f1b4b25ae324560e94b59593a38c0d2e7b6c9872126a9622ed"},
+    {file = "fastrlock-0.8.2-cp39-cp39-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:2587cedbb36c7988e707d83f0f1175c1f882f362b5ebbee25d70218ea33d220d"},
+    {file = "fastrlock-0.8.2-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_24_x86_64.whl", hash = "sha256:9af691a9861027181d4de07ed74f0aee12a9650ac60d0a07f4320bff84b5d95f"},
+    {file = "fastrlock-0.8.2-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:99dd6652bd6f730beadf74ef769d38c6bbd8ee6d1c15c8d138ea680b0594387f"},
+    {file = "fastrlock-0.8.2-cp39-cp39-manylinux_2_5_x86_64.manylinux1_x86_64.whl", hash = "sha256:4d63b6596368dab9e0cc66bf047e7182a56f33b34db141816a4f21f5bf958228"},
+    {file = "fastrlock-0.8.2-cp39-cp39-musllinux_1_1_aarch64.whl", hash = "sha256:ff75c90663d6e8996610d435e71487daa853871ad1770dd83dc0f2fc4997241e"},
+    {file = "fastrlock-0.8.2-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:e27c3cd27fbd25e5223c5c992b300cd4ee8f0a75c6f222ce65838138d853712c"},
+    {file = "fastrlock-0.8.2-cp39-cp39-win_amd64.whl", hash = "sha256:dd961a32a7182c3891cdebca417fda67496d5d5de6ae636962254d22723bdf52"},
+    {file = "fastrlock-0.8.2.tar.gz", hash = "sha256:644ec9215cf9c4df8028d8511379a15d9c1af3e16d80e47f1b6fdc6ba118356a"},
+]
+
 [[package]]
 name = "filelock"
 version = "3.16.1"
@@ -3638,7 +3752,8 @@ test = ["big-O", "importlib-resources", "jaraco.functools", "jaraco.itertools",
 type = ["pytest-mypy"]
 
 [extras]
-all = ["ipywidgets", "nbformat", "nmslib", "nmslib-metabrainz", "plotly", "pytorch-lightning", "rectools-lightfm", "torch", "torch"]
+all = ["cupy-cuda12x", "ipywidgets", "nbformat", "nmslib", "nmslib-metabrainz", "plotly", "pytorch-lightning", "rectools-lightfm", "torch", "torch"]
+cupy = ["cupy-cuda12x"]
 lightfm = ["rectools-lightfm"]
 nmslib = ["nmslib", "nmslib-metabrainz"]
 torch = ["pytorch-lightning", "torch", "torch"]
@@ -3647,4 +3762,4 @@ visuals = ["ipywidgets", "nbformat", "plotly"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.9, <3.13"
-content-hash = "0be90e416bf1a931732bf02749481c4569077fa2ee63291abd6965d604c0d04a"
+content-hash = "894765cd6220c87dd20ad10a0926b39298bfbc2d215ef0091dbf0849a468b349"
diff --git a/pyproject.toml b/pyproject.toml
index 4b13ac67..50a9b973 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -84,6 +84,7 @@ pytorch-lightning = {version = ">=1.6.0, <3.0.0", optional = true}
 ipywidgets = {version = ">=7.7,<8.2", optional = true}
 plotly = {version="^5.22.0", optional = true}
 nbformat = {version = ">=4.2.0", optional = true}
+cupy-cuda12x = {version = "^13.3.0", python = "<3.13", optional = true}
 
 
 [tool.poetry.extras]
@@ -91,11 +92,13 @@ lightfm = ["rectools-lightfm"]
 nmslib = ["nmslib", "nmslib-metabrainz"]
 torch = ["torch", "pytorch-lightning"]
 visuals = ["ipywidgets", "plotly", "nbformat"]
+cupy = ["cupy-cuda12x"]
 all = [
     "rectools-lightfm",
     "nmslib", "nmslib-metabrainz",
     "torch", "pytorch-lightning",
     "ipywidgets", "plotly", "nbformat",
+    "cupy-cuda12x",
 ]
 
 
diff --git a/rectools/models/__init__.py b/rectools/models/__init__.py
index be69f0ce..ac40e522 100644
--- a/rectools/models/__init__.py
+++ b/rectools/models/__init__.py
@@ -28,6 +28,7 @@
 `models.DSSMModel`
 `models.EASEModel`
 `models.ImplicitALSWrapperModel`
+`models.ImplicitBPRWrapperModel`
 `models.ImplicitItemKNNWrapperModel`
 `models.LightFMWrapperModel`
 `models.PopularModel`
@@ -48,7 +49,7 @@
 from .popular_in_category import PopularInCategoryModel
 from .pure_svd import PureSVDModel
 from .random import RandomModel
-from .serialization import load_model, model_from_config
+from .serialization import load_model, model_from_config, model_from_params
 
 try:
     from .lightfm import LightFMWrapperModel
@@ -76,4 +77,5 @@
     "DSSMModel",
     "load_model",
     "model_from_config",
+    "model_from_params",
 )
diff --git a/rectools/models/base.py b/rectools/models/base.py
index c7569918..604c3fb3 100644
--- a/rectools/models/base.py
+++ b/rectools/models/base.py
@@ -31,7 +31,7 @@
 from rectools.exceptions import NotFittedError
 from rectools.types import ExternalIdsArray, InternalIdsArray
 from rectools.utils.config import BaseConfig
-from rectools.utils.misc import get_class_or_function_full_path, import_object, make_dict_flat
+from rectools.utils.misc import get_class_or_function_full_path, import_object, make_dict_flat, unflatten_dict
 from rectools.utils.serialization import PICKLE_PROTOCOL, FileLike, read_bytes
 
 T = tp.TypeVar("T", bound="ModelBase")
@@ -210,6 +210,26 @@ def from_config(cls, config: tp.Union[dict, ModelConfig_T]) -> tpe.Self:
 
         return cls._from_config(config_obj)
 
+    @classmethod
+    def from_params(cls, params: tp.Dict[str, tp.Any], sep: str = ".") -> tpe.Self:
+        """
+        Create model from parameters.
+        Same as `from_config` but accepts flat dict.
+
+        Parameters
+        ----------
+        params : dict
+            Model parameters as a flat dict with keys separated by `sep`.
+        sep : str, default "."
+            Separator for nested keys.
+
+        Returns
+        -------
+        Model instance.
+        """
+        config_dict = unflatten_dict(params, sep=sep)
+        return cls.from_config(config_dict)
+
     @classmethod
     def _from_config(cls, config: ModelConfig_T) -> tpe.Self:
         raise NotImplementedError()
diff --git a/rectools/models/pure_svd.py b/rectools/models/pure_svd.py
index 3d5d5de5..a0ba2153 100644
--- a/rectools/models/pure_svd.py
+++ b/rectools/models/pure_svd.py
@@ -15,6 +15,7 @@
 """SVD Model."""
 
 import typing as tp
+import warnings
 
 import numpy as np
 import typing_extensions as tpe
@@ -26,6 +27,15 @@
 from rectools.models.rank import Distance
 from rectools.models.vector import Factors, VectorModel
 
+try:
+    import cupy as cp
+    from cupyx.scipy.sparse import csr_matrix as cp_csr_matrix
+    from cupyx.scipy.sparse.linalg import svds as cupy_svds
+except ImportError:  # pragma: no cover
+    cupy_svds = None
+    cp_csr_matrix = None
+    cp = None
+
 
 class PureSVDModelConfig(ModelConfig):
     """Config for `PureSVD` model."""
@@ -34,6 +44,7 @@ class PureSVDModelConfig(ModelConfig):
     tol: float = 0
     maxiter: tp.Optional[int] = None
     random_state: tp.Optional[int] = None
+    use_gpu: tp.Optional[bool] = False
     recommend_n_threads: int = 0
     recommend_use_gpu_ranking: bool = True
 
@@ -53,7 +64,9 @@ class PureSVDModel(VectorModel[PureSVDModelConfig]):
     maxiter : int, optional, default ``None``
         Maximum number of iterations.
     random_state : int, optional, default ``None``
-        Pseudorandom number generator state used to generate resamples.
+        Pseudorandom number generator state used to generate resamples. Omitted if use_gpu is True.
+    use_gpu : bool, default ``False``
+        If ``True``, `cupyx.scipy.sparse.linalg.svds()` is used instead of SciPy. CuPy is required.
     verbose : int, default ``0``
         Degree of verbose output. If ``0``, no output will be provided.
     recommend_n_threads: int, default 0
@@ -83,6 +96,7 @@ def __init__(
         tol: float = 0,
         maxiter: tp.Optional[int] = None,
         random_state: tp.Optional[int] = None,
+        use_gpu: tp.Optional[bool] = False,
         verbose: int = 0,
         recommend_n_threads: int = 0,
         recommend_use_gpu_ranking: bool = True,
@@ -93,6 +107,16 @@ def __init__(
         self.tol = tol
         self.maxiter = maxiter
         self.random_state = random_state
+        self._use_gpu = use_gpu  # for making a config
+        if use_gpu:  # pragma: no cover
+            if not cp:
+                warnings.warn("Forced to use CPU. CuPy is not available.")
+                use_gpu = False
+            elif not cp.cuda.is_available():
+                warnings.warn("Forced to use CPU. GPU is not available.")
+                use_gpu = False
+
+        self.use_gpu = use_gpu
         self.recommend_n_threads = recommend_n_threads
         self.recommend_use_gpu_ranking = recommend_use_gpu_ranking
 
@@ -106,6 +130,7 @@ def _get_config(self) -> PureSVDModelConfig:
             tol=self.tol,
             maxiter=self.maxiter,
             random_state=self.random_state,
+            use_gpu=self._use_gpu,
             verbose=self.verbose,
             recommend_n_threads=self.recommend_n_threads,
             recommend_use_gpu_ranking=self.recommend_use_gpu_ranking,
@@ -118,6 +143,7 @@ def _from_config(cls, config: PureSVDModelConfig) -> tpe.Self:
             tol=config.tol,
             maxiter=config.maxiter,
             random_state=config.random_state,
+            use_gpu=config.use_gpu,
             verbose=config.verbose,
             recommend_n_threads=config.recommend_n_threads,
             recommend_use_gpu_ranking=config.recommend_use_gpu_ranking,
@@ -126,10 +152,19 @@ def _from_config(cls, config: PureSVDModelConfig) -> tpe.Self:
     def _fit(self, dataset: Dataset) -> None:  # type: ignore
         ui_csr = dataset.get_user_item_matrix(include_weights=True)
 
-        u, sigma, vt = svds(ui_csr, k=self.factors, tol=self.tol, maxiter=self.maxiter, random_state=self.random_state)
+        if self.use_gpu:  # pragma: no cover
+            ui_csr = cp_csr_matrix(ui_csr)
+            # To prevent IndexError, we need to subtract 1 from factors
+            u, sigma, vt = cupy_svds(ui_csr.toarray(), k=self.factors - 1, tol=self.tol, maxiter=self.maxiter)
+            u = u.get()
+            self.item_factors = (cp.diag(sigma) @ vt).T.get()
+        else:
+            u, sigma, vt = svds(
+                ui_csr, k=self.factors, tol=self.tol, maxiter=self.maxiter, random_state=self.random_state
+            )
+            self.item_factors = (np.diag(sigma) @ vt).T
 
         self.user_factors = u
-        self.item_factors = (np.diag(sigma) @ vt).T
 
     def _get_users_factors(self, dataset: Dataset) -> Factors:
         return Factors(self.user_factors)
diff --git a/rectools/models/serialization.py b/rectools/models/serialization.py
index 91844187..f4ce3c58 100644
--- a/rectools/models/serialization.py
+++ b/rectools/models/serialization.py
@@ -18,6 +18,7 @@
 from pydantic import TypeAdapter
 
 from rectools.models.base import ModelBase, ModelClass, ModelConfig
+from rectools.utils.misc import unflatten_dict
 from rectools.utils.serialization import FileLike, read_bytes
 
 
@@ -46,7 +47,7 @@ def model_from_config(config: tp.Union[dict, ModelConfig]) -> ModelBase:
 
     Parameters
     ----------
-    config : ModelConfig
+    config : dict or ModelConfig
         Model config.
 
     Returns
@@ -64,3 +65,24 @@ def model_from_config(config: tp.Union[dict, ModelConfig]) -> ModelBase:
         raise ValueError("`cls` must be provided in the config to load the model")
 
     return model_cls.from_config(config)
+
+
+def model_from_params(params: dict, sep: str = ".") -> ModelBase:
+    """
+    Create model from dict of parameters.
+    Same as `from_config` but accepts flat dict.
+
+    Parameters
+    ----------
+    params : dict
+        Model parameters as a flat dict with keys separated by `sep`.
+    sep : str, default "."
+        Separator for nested keys.
+
+    Returns
+    -------
+    model
+        Model instance.
+    """
+    config_dict = unflatten_dict(params, sep=sep)
+    return model_from_config(config_dict)
diff --git a/rectools/utils/misc.py b/rectools/utils/misc.py
index 3e6ba433..1d81da14 100644
--- a/rectools/utils/misc.py
+++ b/rectools/utils/misc.py
@@ -228,3 +228,34 @@ def make_dict_flat(d: tp.Dict[str, tp.Any], sep: str = ".", parent_key: str = ""
         else:
             items.append((new_key, v))
     return dict(items)
+
+
+def unflatten_dict(d: tp.Dict[str, tp.Any], sep: str = ".") -> tp.Dict[str, tp.Any]:
+    """
+    Convert a flat dict with concatenated keys back into a nested dictionary.
+
+    Parameters
+    ----------
+    d : dict
+        Flattened dictionary.
+    sep : str, default "."
+        Separator used in flattened keys.
+
+    Returns
+    -------
+    dict
+        Nested dictionary.
+
+    Examples
+    --------
+    >>> unflatten_dict({'a.b': 1, 'a.c': 2, 'd': 3})
+    {'a': {'b': 1, 'c': 2}, 'd': 3}
+    """
+    result: tp.Dict[str, tp.Any] = {}
+    for key, value in d.items():
+        parts = key.split(sep)
+        current = result
+        for part in parts[:-1]:
+            current = current.setdefault(part, {})
+        current[parts[-1]] = value
+    return result
diff --git a/tests/models/test_base.py b/tests/models/test_base.py
index 5ab7d68e..d281017f 100644
--- a/tests/models/test_base.py
+++ b/tests/models/test_base.py
@@ -19,6 +19,7 @@
 from datetime import timedelta
 from pathlib import Path
 from tempfile import NamedTemporaryFile, TemporaryFile
+from unittest.mock import MagicMock
 
 import numpy as np
 import pandas as pd
@@ -498,6 +499,15 @@ def test_from_config_dict_with_extra_keys(self) -> None:
         ):
             self.model_class.from_config(config)
 
+    def test_from_params(self, mocker: MagicMock) -> None:
+        params = {"x": 10, "verbose": 1, "sc.td": "P2DT3H"}
+        spy = mocker.spy(self.model_class, "from_config")
+        model = self.model_class.from_params(params)
+        spy.assert_called_once_with({"x": 10, "verbose": 1, "sc": {"td": "P2DT3H"}})
+        assert model.x == 10
+        assert model.td == timedelta(days=2, hours=3)
+        assert model.verbose == 1
+
     def test_get_config_pydantic(self) -> None:
         model = self.model_class(x=10, verbose=1)
         config = model.get_config(mode="pydantic")
diff --git a/tests/models/test_pure_svd.py b/tests/models/test_pure_svd.py
index 71a7ed67..408023da 100644
--- a/tests/models/test_pure_svd.py
+++ b/tests/models/test_pure_svd.py
@@ -17,6 +17,7 @@
 import numpy as np
 import pandas as pd
 import pytest
+from pytest_mock import MockerFixture
 
 from rectools import Columns
 from rectools.dataset import Dataset
@@ -32,6 +33,17 @@
     assert_second_fit_refits_model,
 )
 
+try:
+    import cupy as cp  # pylint: disable=import-error, unused-import
+except ImportError:  # pragma: no cover
+    cp = None
+
+try:
+    HAS_CUDA = cp.is_available() if cp else False
+except Exception:  # pragma: no cover  # pylint: disable=broad-except
+    # If CUDA isn't installed cupy raises CUDARuntimeError:
+    HAS_CUDA = False
+
 
 class TestPureSVDModel:
 
@@ -72,6 +84,7 @@ def test_basic(
         expected: pd.DataFrame,
         use_gpu_ranking: bool,
     ) -> None:
+
         model = PureSVDModel(factors=2, recommend_use_gpu_ranking=use_gpu_ranking).fit(dataset)
         actual = model.recommend(
             users=np.array([10, 20]),
@@ -85,6 +98,52 @@ def test_basic(
             actual,
         )
 
+    # SciPy's svds and cupy's svds results can be different and use_gpu fallback causes errors
+    @pytest.mark.skipif(cp is None or not HAS_CUDA, reason="CUDA is not available")
+    @pytest.mark.parametrize(
+        "filter_viewed,expected",
+        (
+            (
+                True,
+                pd.DataFrame(
+                    {
+                        Columns.User: [10, 10, 20, 20],
+                        Columns.Item: [15, 13, 14, 15],
+                        Columns.Rank: [1, 2, 1, 2],
+                    }
+                ),
+            ),
+            (
+                False,
+                pd.DataFrame(
+                    {
+                        Columns.User: [10, 10, 20, 20],
+                        Columns.Item: [11, 12, 11, 12],
+                        Columns.Rank: [1, 2, 1, 2],
+                    }
+                ),
+            ),
+        ),
+    )
+    def test_basic_gpu(
+        self,
+        dataset: Dataset,
+        filter_viewed: bool,
+        expected: pd.DataFrame,
+    ) -> None:
+        model = PureSVDModel(factors=2, use_gpu=True, recommend_use_gpu_ranking=True).fit(dataset)
+        actual = model.recommend(
+            users=np.array([10, 20]),
+            dataset=dataset,
+            k=2,
+            filter_viewed=filter_viewed,
+        )
+        pd.testing.assert_frame_equal(actual.drop(columns=Columns.Score), expected)
+        pd.testing.assert_frame_equal(
+            actual.sort_values([Columns.User, Columns.Score], ascending=[True, False]).reset_index(drop=True),
+            actual,
+        )
+
     @pytest.mark.parametrize(
         "filter_viewed,expected",
         (
@@ -279,12 +338,16 @@ def test_dumps_loads(self, dataset: Dataset) -> None:
 
 class TestPureSVDModelConfiguration:
 
-    def test_from_config(self) -> None:
+    @pytest.mark.parametrize("use_gpu", (False, True))
+    def test_from_config(self, mocker: MockerFixture, use_gpu: bool) -> None:
+        mocker.patch("rectools.models.pure_svd.cp", return_value=True)
+        mocker.patch("rectools.models.pure_svd.cp.cuda.is_available", return_value=True)
         config = {
             "factors": 100,
             "tol": 0,
             "maxiter": 100,
             "random_state": 32,
+            "use_gpu": use_gpu,
             "verbose": 0,
         }
         model = PureSVDModel.from_config(config)
@@ -296,12 +359,18 @@ def test_from_config(self) -> None:
 
     @pytest.mark.parametrize("random_state", (None, 42))
     @pytest.mark.parametrize("simple_types", (False, True))
-    def test_get_config(self, random_state: tp.Optional[int], simple_types: bool) -> None:
+    @pytest.mark.parametrize("use_gpu", (False, True))
+    def test_get_config(
+        self, mocker: MockerFixture, random_state: tp.Optional[int], simple_types: bool, use_gpu: bool
+    ) -> None:
+        mocker.patch("rectools.models.pure_svd.cp.cuda.is_available", return_value=True)
+        mocker.patch("rectools.models.pure_svd.cp", return_value=True)
         model = PureSVDModel(
             factors=100,
-            tol=1,
+            tol=1.0,
             maxiter=100,
             random_state=random_state,
+            use_gpu=use_gpu,
             verbose=1,
             recommend_n_threads=2,
             recommend_use_gpu_ranking=False,
@@ -310,9 +379,10 @@ def test_get_config(self, random_state: tp.Optional[int], simple_types: bool) ->
         expected = {
             "cls": "PureSVDModel" if simple_types else PureSVDModel,
             "factors": 100,
-            "tol": 1,
+            "tol": 1.0,
             "maxiter": 100,
             "random_state": random_state,
+            "use_gpu": use_gpu,
             "verbose": 1,
             "recommend_n_threads": 2,
             "recommend_use_gpu_ranking": False,
diff --git a/tests/models/test_serialization.py b/tests/models/test_serialization.py
index ce95af17..f3340707 100644
--- a/tests/models/test_serialization.py
+++ b/tests/models/test_serialization.py
@@ -14,6 +14,7 @@
 
 import typing as tp
 from tempfile import NamedTemporaryFile
+from unittest.mock import MagicMock
 
 import pytest
 from implicit.als import AlternatingLeastSquares
@@ -26,7 +27,6 @@
 except ImportError:
     LightFM = object  # it's ok in case we're skipping the tests
 
-
 from rectools.metrics import NDCG
 from rectools.models import (
     DSSMModel,
@@ -39,10 +39,13 @@
     PopularModel,
     load_model,
     model_from_config,
+    model_from_params,
+    serialization,
 )
 from rectools.models.base import ModelBase, ModelConfig
 from rectools.models.nn.transformer_base import TransformerModelBase
 from rectools.models.vector import VectorModel
+from rectools.utils.config import BaseConfig
 
 from .utils import get_successors
 
@@ -79,20 +82,26 @@ def test_load_model(model_cls: tp.Type[ModelBase]) -> None:
     assert not loaded_model.is_fitted
 
 
+class CustomModelSubConfig(BaseConfig):
+    x: int = 10
+
+
 class CustomModelConfig(ModelConfig):
     some_param: int = 1
+    sc: CustomModelSubConfig = CustomModelSubConfig()
 
 
 class CustomModel(ModelBase[CustomModelConfig]):
     config_class = CustomModelConfig
 
-    def __init__(self, some_param: int = 1, verbose: int = 0):
+    def __init__(self, some_param: int = 1, x: int = 10, verbose: int = 0):
         super().__init__(verbose=verbose)
         self.some_param = some_param
+        self.x = x
 
     @classmethod
     def _from_config(cls, config: CustomModelConfig) -> "CustomModel":
-        return cls(some_param=config.some_param, verbose=config.verbose)
+        return cls(some_param=config.some_param, x=config.sc.x, verbose=config.verbose)
 
 
 class TestModelFromConfig:
@@ -121,6 +130,7 @@ def test_custom_model_creation(self, config: tp.Union[dict, CustomModelConfig])
         model = model_from_config(config)
         assert isinstance(model, CustomModel)
         assert model.some_param == 2
+        assert model.x == 10
 
     @pytest.mark.parametrize("simple_types", (False, True))
     def test_fails_on_missing_cls(self, simple_types: bool) -> None:
@@ -179,3 +189,15 @@ def test_fails_on_model_cls_without_from_config_support(self, model_cls: tp.Any)
         config = {"cls": model_cls}
         with pytest.raises(NotImplementedError, match="`from_config` method is not implemented for `DSSMModel` model"):
             model_from_config(config)
+
+
+class TestModelFromParams:
+    def test_uses_from_config(self, mocker: MagicMock) -> None:
+        params = {"cls": "tests.models.test_serialization.CustomModel", "some_param": 2, "sc.x": 20}
+        spy = mocker.spy(serialization, "model_from_config")
+        model = model_from_params(params)
+        expected_config = {"cls": "tests.models.test_serialization.CustomModel", "some_param": 2, "sc": {"x": 20}}
+        spy.assert_called_once_with(expected_config)
+        assert isinstance(model, CustomModel)
+        assert model.some_param == 2
+        assert model.x == 20
diff --git a/tests/utils/test_misc.py b/tests/utils/test_misc.py
new file mode 100644
index 00000000..eb493f7e
--- /dev/null
+++ b/tests/utils/test_misc.py
@@ -0,0 +1,42 @@
+from rectools.utils.misc import unflatten_dict
+
+
+class TestUnflattenDict:
+    def test_empty(self) -> None:
+        assert unflatten_dict({}) == {}
+
+    def test_complex(self) -> None:
+        flattened = {
+            "a.b": 1,
+            "a.c": 2,
+            "d": 3,
+            "a.e.f": [10, 20],
+        }
+        excepted = {
+            "a": {"b": 1, "c": 2, "e": {"f": [10, 20]}},
+            "d": 3,
+        }
+        assert unflatten_dict(flattened) == excepted
+
+    def test_simple(self) -> None:
+        flattened = {
+            "a": 1,
+            "b": 2,
+        }
+        excepted = {
+            "a": 1,
+            "b": 2,
+        }
+        assert unflatten_dict(flattened) == excepted
+
+    def test_non_default_sep(self) -> None:
+        flattened = {
+            "a_b": 1,
+            "a_c": 2,
+            "d": 3,
+        }
+        excepted = {
+            "a": {"b": 1, "c": 2},
+            "d": 3,
+        }
+        assert unflatten_dict(flattened, sep="_") == excepted