52e20f2daf
Update conversion script args
2023-05-31 15:13:19 +02:00
eed6a05c5f
Update supported_models.py
2023-05-31 15:07:54 +02:00
84719a5fa3
Add `--skip_validation` option to conversion script
2023-05-31 11:59:20 +02:00
bcaf7dd155
Remove torch_dtype from config
2023-05-31 02:28:16 +02:00
99c771bf48
Run `generate_tests.py` as module
2023-05-31 02:21:15 +02:00
813b2abe77
Remove unnecessary imports
2023-05-31 02:06:34 +02:00
e3f73cd17d
Add unit tests for config files
2023-05-31 02:04:58 +02:00
8a4dcbf2ef
Add JSDoc to imported Tensor class
2023-05-31 02:04:27 +02:00
3d05083c1a
Add config auto class
2023-05-31 02:04:07 +02:00
44caca829f
Fix JSDoc for tokenizers
2023-05-31 01:43:50 +02:00
db2bcdfd55
Fix masked LM unit tests
2023-05-31 01:43:31 +02:00
40149ebb6b
Update supported models
2023-05-30 19:37:34 +02:00
5e53f2cf02
Rename tasks script
2023-05-30 17:13:27 +02:00
ec00d4f540
Update default quantization params in conversion script
2023-05-30 14:57:44 +02:00
55d6ef41b3
[docs] Fix numbering
2023-05-29 17:49:46 +02:00
80d92cc38b
Use error mapping instead of switch block
2023-05-29 17:23:39 +02:00
4feb8b4527
Correctly use default module if present
2023-05-29 17:21:19 +02:00
14c02f8760
Add `sentence-transformers` models to supported models/tasks
2023-05-29 17:19:33 +02:00
331c8700ae
Replace `Math.max` with custom `max` function
2023-05-29 16:51:38 +02:00
dad26f1c0a
Update README.md
2023-05-29 15:46:44 +02:00
8b89766034
Update package-lock.json
2023-05-29 12:51:15 +02:00
9bc1c33b39
[version] Update to 2.0.2
2023-05-29 12:50:30 +02:00
2f2092263d
Use versioned links (#Closes #114 )
...
Prevents issues where users copy-paste the import code, and then a future update breaks it.
Also ensures that the default wasm files match the target version
2023-05-29 12:50:17 +02:00
ac23bc5e12
Freeze onnxruntime dependencies ( #124 )
...
Their latest version has a few issues, particularly with webgpu, and also uses .wasm files which are incompatible with their previous versions.
So, while those issues are sorted out, it's best to freeze their packages to the latest stable version.
2023-05-29 12:41:52 +02:00
efb45ba392
Improve cache key
...
- When accessing remote file, use the url as the cache key
- Also use sensible cache key when running locally (with access to file system)
2023-05-28 23:33:46 +02:00
d6ca1eaa90
Remove `cos_sim` function from pipeline
...
The user should use the `cos_sim` or `dot` functions in utils/maths.js
2023-05-28 18:46:34 +02:00
7c041d84b4
Update feature extraction test cases
2023-05-28 18:45:45 +02:00
33b37061c3
Improve feature extraction pipeline
...
Now works for models that are not specifically used for feature extraction
2023-05-26 12:06:43 +02:00
85f76b3285
Use generated request ID for caching model files
...
Fixes issue when requesting different revisions or quantizations of models
2023-05-26 10:57:41 +02:00
3edbbdaffb
Implement tensor ops
...
- `sum`
- `norm`
- `normalize`
- `stride`
- `squeeze`
- `unsqueeze`
- `mean_pooling`
2023-05-26 10:49:51 +02:00
d7c353d812
[version] Update to 2.0.1
2023-05-20 15:47:07 +02:00
63240244fc
[package.json] Update keywords
2023-05-20 15:46:44 +02:00
e71a2eaa16
Build demo website after release
2023-05-20 15:37:36 +02:00
1e2d1e2eb4
Add tokenizer test cases with numbers
2023-05-19 10:40:54 +02:00
11ae173957
Update Bert pretokenizer regex
...
Should match the rust implementation: b4fcc9ce6e/tokenizers/src/pre_tokenizers/bert.rs (L11)
2023-05-19 10:39:08 +02:00
eaf8f6ba6c
Fix Bert tokenizer regex for numbers
2023-05-19 08:49:50 +02:00
4a6b8ccc41
[demo] Fix token-classification ( Closes #116 )
2023-05-19 08:04:45 +02:00
8d8f511b34
Update zero-shot image classification test case
2023-05-17 18:40:35 +02:00
112420174b
Release version 2.0.0
2023-05-17 15:14:09 +02:00
75ec68ed8b
Create example Node.js application
2023-05-17 12:34:48 +02:00
20b6e6d2e4
Update version to 2.0.0-alpha.4
2023-05-17 00:04:04 +02:00
a53da6bf4a
Rename `CustomImage` to `RawImage`
2023-05-16 23:02:26 +02:00
815a77dd78
Create test suite for utility functions
2023-05-16 22:35:57 +02:00
7ac9dc71c0
Update processors.js
2023-05-16 22:35:04 +02:00
acc53dd8b1
Improve `getMelFilters` function
...
Minor performance +memory improvements (~20%)
2023-05-16 22:28:38 +02:00
32cb92d073
Only calculate mel filters once (during initialization)
2023-05-16 20:36:36 +02:00
f424353c61
Merge branch 'pr/112'
2023-05-16 20:06:32 +02:00
0b403ce8e4
Set base path for demo site from environment variable if present
2023-05-16 19:10:02 +02:00
be907015e4
Update gh-pages.yml
2023-05-16 19:01:11 +02:00
56e0c06580
Update package-lock.json
2023-05-16 19:00:28 +02:00
Joshua Lochner