upstream-archive/fastembed
1
0
Fork 0
mirror of https://github.com/qdrant/fastembed.git synced 2026-08-03 16:41:00 -05:00
Code Issues Packages Projects Releases Wiki Activity
250 Commits 38 Branches 59 Tags
push
Commit Graph

250 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
NirantK
7fdd48c3e2 * fix(embedding.py): update NotImplementedError message to provide more specific information 
* fix(embedding.py): update ValueError message to provide more specific reasons for the error
0.0.5a1 push 		2023-09-25 17:40:55 +05:30
NirantK
547130a4c3 Move throughput comparison to docs 2023-09-25 17:40:38 +05:30
NirantK
f51afec563 * chore(README.md): remove unnecessary section heading "Under the hood" 
* docs(README.md): update bullet point description in "Why fast?" section
* docs(README.md): update bullet point description in "Why light?" section
* docs(README.md): update bullet point description in
 2023-09-25 17:32:04 +05:30
NirantK
250ccf98aa * docs(README.md): add section on similar work and reference to Ilyas M.'s tweet about using FlagEmbeddings with Optimum over CUDA 2023-09-25 17:30:56 +05:30
NirantK
4da2be9c44 Remove the else, it's cleaner 2023-09-25 17:15:44 +05:30
NirantK
206a633eb1 Fix for # https://github.com/pytorch/pytorch/issues/100974 2023-09-25 17:12:03 +05:30
NirantK
c0787c0aeb * fix(embedding.py): replace torch.nn.functional.normalize with custom normalize function 2023-09-25 16:59:37 +05:30
NirantK
5dfdddbdb6 Another attempt at fixing versions 2023-09-25 16:53:14 +05:30
NirantK
4d6d27cffb * test(test_onnx_embeddings.py): remove unnecessary list conversion in embeddings assignment 2023-09-25 16:50:52 +05:30
NirantK
54a88d0028 update lock 2023-09-25 16:48:23 +05:30
NirantK
6810141e87 * fix(embedding.py): fix model_name splitting to correctly extract the model name 
* fix(embedding.py): handle PermissionError when downloading fast_model_name.tar.gz and try simple_model_name.tar.gz as a fallback
* fix(embedding.py): raise ValueError if neither fast_model_name.tar.gz nor simple are valid
 2023-09-25 16:45:39 +05:30
NirantK
464b12ab15 * chore(pyproject.toml): update optimum dependency version to be greater than 1.12.0 2023-09-25 16:14:24 +05:30
NirantK
633fbde726 * fix(embedding.py): rename file model_optimized.onnx to model.onnx if it exists 
* fix(embedding.py): revert back to the default model to avoid confusion
 2023-09-25 16:05:22 +05:30
NirantK
3cf17b8707 Rename nbs to fooling_around 2023-09-25 15:56:41 +05:30
NirantK
e1c1792532 * refactor(embedding.py): remove unused imports and variables 2023-09-25 15:56:25 +05:30
NirantK
24be1d946f update the default embedding model 2023-09-25 15:55:27 +05:30
NirantK
3149694c04 Utility script version of the notebook 2023-09-25 15:33:44 +05:30
NirantK
f007b19272 Add optimum as a dependency 2023-09-25 15:33:28 +05:30
NirantK
9c3fd66adf Replace ONNXRuntime with Optimum Usage 2023-09-25 15:33:18 +05:30
NirantK
eaafd59277 wip 2023-09-18 22:29:39 +05:30
NirantK
34f9982af3 add benchmarking 2023-09-18 18:32:50 +05:30
NirantK
0ab82da4aa * chore(.gitignore): add qdrant_storage/ directory to the gitignore file 2023-09-18 17:24:31 +05:30
NirantK
6f284f6466 * feat(docs): update execution counts in Binary Quantization from Scratch.ipynb 
* fix(docs): fix typo in Binary Quantization from Scratch.ipynb
* feat(docs): add outputs to code cells in Binary Quantization from Scratch.ipynb
 2023-09-18 08:47:28 +05:30
NirantK
ac989e249a Add BQ with Qdrant draft 2023-09-18 08:47:11 +05:30
NirantK
5906ef63d6 * chore(.gitignore): add ignore rule for experimental .bin files in docs/experimental directory 2023-09-18 08:46:58 +05:30
NirantK
b72cae1e9b * chore(docs): rename HF_vs_FastEmbed.ipynb to 02_HF_vs_FastEmbed.ipynb 2023-09-18 08:46:13 +05:30
NirantK
ecfaba9e60 Fix yield/iterator issue 2023-09-18 08:42:34 +05:30
NirantK
fc0341333e * chore(docs): remove experimental notebook for 1M Embedding Creation 2023-09-18 08:41:58 +05:30
NirantK
3023996074 * fix(embedding.py): add support for ONNX runtime session options 2023-09-18 08:30:44 +05:30
NirantK
296346a1a9 * chore(embedding.py): refactor FlagEmbedding class to use separate method for ONNX embedding 
* feat(embedding.py): add support for ONNXProviders.Metal in FlagEmbedding constructor
 2023-09-18 08:08:20 +05:30
NirantK
a335c8898f Remove attention pooling 2023-09-18 07:45:12 +05:30
NirantK
5afa103d45 Remove attention pooling 2023-09-18 07:35:52 +05:30
NirantK
e1e5975193 * docs(experimental): rename Binary Quantization.ipynb to Binary Quantization from Scratch.ipynb 2023-09-15 15:33:28 +05:30
NirantK
a7a599d932 * chore(main.html): update Discord server link and name to Qdrant Discord server 2023-09-13 20:53:21 +05:30
NirantK
9c5d32f271 move nbs 2023-09-07 15:25:32 +05:30
Nirant
defbe0cce2 Merge pull request #8 from qdrant/add_qdrant_client_example 
Add Qdrant Usage back to the example
 2023-09-07 15:22:43 +05:30
NirantK
1292ded017 Add Qdrant Usage back to the example 2023-09-07 15:19:46 +05:30
Nirant
9cd13a890e Merge pull request #7 from Rishav-hub/patch-1 
Updated the hyperlink for "Retrieval with FastEmbed"
 2023-09-07 15:14:01 +05:30
Rishav Dash
bcf3cee8b8 Updated the hyperlink for "Retrieval with FastEmbed" 
prev -: https://qdrant.github.io/fastembed/examples/Retrieval%20with%20FastEmbed/

now -: https://qdrant.github.io/fastembed/examples/Retrieval_with_FastEmbed/
 2023-09-03 13:37:14 +05:30
NirantK
728ac49a91 * docs(experimental): update headings in Binary Quantization.ipyn, add emoji everywhere 2023-08-25 14:11:03 +05:30
NirantK
3d1b80e9de * docs(experimental): fix typo in Binary Quantization.ipynb 
* chore(experimental): update description of binary transformation step
 2023-08-25 14:06:16 +05:30
NirantK
86e3b12e84 Improved readability 2023-08-25 14:05:21 +05:30
NirantK
e62ed5edb2 * chore(docs): rename "Binary Quantisation.ipynb" to "Binary Quantization.ipynb" 
* docs(experimental/Binary Quantization.ipynb): update markdown headings
* docs(experimental/Binary Quantization.ipynb): update markdown heading for loading data
 2023-08-25 13:58:45 +05:30
NirantK
ee87685fa5 * docs(experimental): add table with sampling rate, limit, and accuracy for binary quantisation 2023-08-25 13:54:36 +05:30
NirantK
750c9b340e * chore(Binary Quantisation.ipynb): update execution counts to null 
* refactor(Binary Quantisation.ipynb): remove unnecessary outputs
 2023-08-25 13:54:31 +05:30
NirantK
5a16ce8c63 * chore(docs): rename notebook files 
* - Rename '03_HF_vs_FastEmbed.ipynb' to 'HF_vs_FastEmbed.ipynb'
* - Rename '01_Retrieval_with_FastEmbed.ipynb' to 'Retrieval_with_FastEmbed
 2023-08-25 13:41:32 +05:30
NirantK
5f229b2bc3 Binary Quantisation 2023-08-25 13:40:23 +05:30
NirantK
743a095a93 * chore(docs): update HF_vs_FastEmbed.ipynb and 04_1M_Embedding_Creation.ipynb 
*
* HF_vs_FastEmbed.ipynb:
* - Add code to calculate embeddings using mean pooling with attention weighting
* - Fix variable name error in line 4
*
*
 2023-08-25 12:59:22 +05:30
NirantK
1bd639b817 * chore(docs): rename 02_Usage_with_Qdrant.ipynb to FastEmbed_Usage_with_Qdrant.ipynb in the experimental folder 2023-08-25 12:56:38 +05:30
NirantK
399d11eb8a * chore(.gitignore): add rule to ignore experimental .parquet files in docs/experimental directory 2023-08-25 12:56:25 +05:30