The clip_similarwords is the implementation of finding similar 1-token words of OpenAI's [CLIP](https://github.com/openai/CLIP) in less than one second.
OpenAI's CLIP uses text-image similarities so its text-text similarities may also be text's typical image similarities unlike [WordNet](https://en.wikipedia.org/wiki/WordNet) or other synonym dictionaries.
Note that, for speed and storage reason (PyPI is limited to 60MB), the words composed by 2 or more tokens are not supported.
Installation
============
clip_similarwords is easily installable via pip command:
```bash
pip install clip_similarwords
```
or
```bash
pip install git+https://github.com/nazodane/clip_similarwords.git
```
Usage of the command
====================
```bash
~/.local/bin/clip-similarwords [ word_fragment | --all ]
```
Usage of the module
===================
```python
from clip_similarwords import CLIPTextSimilarWords
clipsim = CLIPTextSimilarWords()
for key_token, sim_token, cos_similarity in clipsim("cat"):
print("%s -> %s ( cos_similarity: %.2f )"%(key_token, sim_token, cos_similarity))
```
Requirements for model uses
===========================
* Linux (should also works on other environmets)
no PyTorch nor CUDA are required.
Requirements for model generation
=================================
* Linux
* Python 3.10 or later
* PyTorch 1.13 or later
* CUDA 11.7 or later
* DRAM 16GB or higher
* RTX 3060 12GB or higher
The patches and informations on other enviroments are surely welcome!
License
=======
The codes are under MIT License. The model was converted under Japanese law.
Raw data
{
"_id": null,
"home_page": "https://github.com/nazodane/clip_similarwords",
"name": "clip-similarwords",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.10.0",
"maintainer_email": "",
"keywords": "clip",
"author": "Toshimitsu Kimura",
"author_email": "lovesyao@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/c0/cb/2dd0e347be71e2f88c076203a24f002773275d8cf1bb6e8960a2469cd6ba/clip_similarwords-0.0.4.1.tar.gz",
"platform": null,
"description": "The clip_similarwords is the implementation of finding similar 1-token words of OpenAI's [CLIP](https://github.com/openai/CLIP) in less than one second. \n\nOpenAI's CLIP uses text-image similarities so its text-text similarities may also be text's typical image similarities unlike [WordNet](https://en.wikipedia.org/wiki/WordNet) or other synonym dictionaries.\n\nNote that, for speed and storage reason (PyPI is limited to 60MB), the words composed by 2 or more tokens are not supported. \n\nInstallation\n============\nclip_similarwords is easily installable via pip command:\n```bash\npip install clip_similarwords\n```\nor\n```bash\npip install git+https://github.com/nazodane/clip_similarwords.git\n```\n\nUsage of the command\n====================\n```bash\n~/.local/bin/clip-similarwords [ word_fragment | --all ]\n```\n\nUsage of the module\n===================\n```python\nfrom clip_similarwords import CLIPTextSimilarWords\nclipsim = CLIPTextSimilarWords()\nfor key_token, sim_token, cos_similarity in clipsim(\"cat\"):\n print(\"%s -> %s ( cos_similarity: %.2f )\"%(key_token, sim_token, cos_similarity))\n```\n\nRequirements for model uses\n===========================\n* Linux (should also works on other environmets)\n\nno PyTorch nor CUDA are required.\n\nRequirements for model generation\n=================================\n* Linux\n* Python 3.10 or later\n* PyTorch 1.13 or later\n* CUDA 11.7 or later\n* DRAM 16GB or higher\n* RTX 3060 12GB or higher\n\nThe patches and informations on other enviroments are surely welcome!\n\nLicense\n=======\nThe codes are under MIT License. The model was converted under Japanese law.\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "finding similar 1-token words on OpenAI's CLIP.",
"version": "0.0.4.1",
"split_keywords": [
"clip"
],
"urls": [
{
"comment_text": "",
"digests": {
"md5": "4648d43f8619ae4775542e0de4bd14d1",
"sha256": "ba70a5003c1d547489846d371442e445c1a2355d809cfda67689cdaff85cfb87"
},
"downloads": -1,
"filename": "clip_similarwords-0.0.4.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "4648d43f8619ae4775542e0de4bd14d1",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.10.0",
"size": 8248362,
"upload_time": "2022-12-22T16:26:57",
"upload_time_iso_8601": "2022-12-22T16:26:57.351990Z",
"url": "https://files.pythonhosted.org/packages/f3/24/076e9bf05d4030e97b2b28c280fcac714237e70d1ab57f80293a20e7a82a/clip_similarwords-0.0.4.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"md5": "267f72fe2c541671b36ce4bf8f1185e2",
"sha256": "ee5868804402b0c2708ef323b704e0861b50c88e6c546922aa8fef5c983e39a7"
},
"downloads": -1,
"filename": "clip_similarwords-0.0.4.1.tar.gz",
"has_sig": false,
"md5_digest": "267f72fe2c541671b36ce4bf8f1185e2",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.10.0",
"size": 8034931,
"upload_time": "2022-12-22T16:27:11",
"upload_time_iso_8601": "2022-12-22T16:27:11.751973Z",
"url": "https://files.pythonhosted.org/packages/c0/cb/2dd0e347be71e2f88c076203a24f002773275d8cf1bb6e8960a2469cd6ba/clip_similarwords-0.0.4.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2022-12-22 16:27:11",
"github": true,
"gitlab": false,
"bitbucket": false,
"github_user": "nazodane",
"github_project": "clip_similarwords",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "clip-similarwords"
}
JSON
