9b17e89 | Patrice Lopez | 04 February 2024, 18:55:15 UTC | update models and scores for copyrights/license | 04 February 2024, 18:55:15 UTC |
2e6b159 | Patrice Lopez | 28 January 2024, 16:24:03 UTC | finish classify option | 28 January 2024, 16:24:03 UTC |
bf19c65 | Patrice Lopez | 28 January 2024, 12:10:31 UTC | more debug | 28 January 2024, 12:10:31 UTC |
7061ff9 | Patrice Lopez | 28 January 2024, 12:03:50 UTC | debug binary classifier data preparation | 28 January 2024, 12:03:50 UTC |
8641e73 | Patrice Lopez | 28 January 2024, 11:22:31 UTC | reference command lines in benchmarks | 28 January 2024, 11:22:31 UTC |
86858eb | Patrice Lopez | 28 January 2024, 10:59:35 UTC | add licenses/copyrights classifier | 28 January 2024, 10:59:35 UTC |
5e9d005 | Patrice Lopez | 25 January 2024, 16:56:26 UTC | resources for updated affiliation-address model | 25 January 2024, 16:56:26 UTC |
e3b0506 | Patrice Lopez | 29 November 2023, 18:45:11 UTC | support local model with safetensors format | 29 November 2023, 18:45:11 UTC |
4d3000e | Patrice Lopez | 29 November 2023, 16:30:04 UTC | release version 0.3.4 | 29 November 2023, 16:30:04 UTC |
236c2e3 | Patrice Lopez | 29 November 2023, 16:26:08 UTC | Merge pull request #165 from kermitt2/max-epoch-param add parameter --max-epoch | 29 November 2023, 16:26:08 UTC |
194767d | Patrice Lopez | 29 November 2023, 16:21:13 UTC | rephrase improving versus changing | 29 November 2023, 16:21:13 UTC |
f4c9d89 | Patrice Lopez | 28 November 2023, 11:14:38 UTC | fix conflict | 28 November 2023, 11:14:38 UTC |
3671cd4 | Patrice Lopez | 28 November 2023, 10:10:30 UTC | Merge branch 'master' of github.com:kermitt2/delft | 28 November 2023, 10:10:30 UTC |
e6d2a52 | Patrice Lopez | 28 November 2023, 10:09:48 UTC | update doc | 28 November 2023, 10:09:48 UTC |
06b0df2 | Patrice Lopez | 28 November 2023, 10:09:44 UTC | update | 28 November 2023, 10:09:44 UTC |
3e68dd9 | Patrice Lopez | 28 November 2023, 10:00:39 UTC | fix conflict | 28 November 2023, 10:00:39 UTC |
f8adbf5 | Patrice Lopez | 28 November 2023, 09:45:38 UTC | Merge pull request #164 from kermitt2/multi-gpu Training with multiple GPUs | 28 November 2023, 09:45:38 UTC |
13c543a | Patrice Lopez | 28 November 2023, 08:42:25 UTC | fix conflict | 28 November 2023, 08:42:25 UTC |
7aceb0d | Patrice Lopez | 27 November 2023, 18:00:59 UTC | Merge pull request #167 from kermitt2/new-models-and-formats New models and formats | 27 November 2023, 18:00:59 UTC |
5e66dcf | Patrice Lopez | 27 November 2023, 17:55:21 UTC | keep current pytorch version | 27 November 2023, 17:55:21 UTC |
7f4a456 | Patrice Lopez | 27 November 2023, 17:43:00 UTC | Merge branch 'new-models-and-formats' of github.com:kermitt2/delft into new-models-and-formats | 27 November 2023, 17:43:00 UTC |
f5f7edc | Patrice Lopez | 27 November 2023, 17:42:48 UTC | fix for dataseer corpus format | 27 November 2023, 17:42:48 UTC |
67c8a63 | Patrice Lopez | 27 November 2023, 17:41:23 UTC | Merge branch 'master' into new-models-and-formats | 27 November 2023, 17:41:23 UTC |
46f809a | Patrice Lopez | 27 November 2023, 17:36:46 UTC | support safetensors HuggingFace model weights format and HF access tokens for loading private models | 27 November 2023, 17:36:46 UTC |
f0da55c | Patrice Lopez | 27 November 2023, 17:35:32 UTC | add funding-acknowledgement support and models | 27 November 2023, 17:35:32 UTC |
379fba1 | Patrice Lopez | 27 November 2023, 17:32:52 UTC | support learning rate command line parameter | 27 November 2023, 17:32:52 UTC |
16f43fd | Patrice Lopez | 27 November 2023, 17:32:02 UTC | update dependencies | 27 November 2023, 17:32:02 UTC |
f8df013 | Luca Foppiano | 01 September 2023, 03:59:09 UTC | fix the closing pool only for tf < 2.10.0 | 01 September 2023, 03:59:09 UTC |
8a03944 | Luca Foppiano | 01 September 2023, 03:57:58 UTC | fix missing return | 01 September 2023, 03:57:58 UTC |
0b066ad | Luca Foppiano | 01 September 2023, 03:50:01 UTC | fix wrong method name call | 01 September 2023, 03:50:01 UTC |
d05ed04 | Luca Foppiano | 01 September 2023, 03:33:27 UTC | enable multi-gpu for inference in sequence labelling | 01 September 2023, 03:45:00 UTC |
e169867 | Luca Foppiano | 01 September 2023, 03:26:25 UTC | fix exception when multi-gpu is used on a single gpu, improve message | 01 September 2023, 03:26:25 UTC |
b3ba6e8 | Patrice Lopez | 31 August 2023, 21:40:28 UTC | Merge pull request #153 from kermitt2/dependabot/pip/torch-1.13.1 Bump torch from 1.10.1 to 1.13.1 | 31 August 2023, 21:40:28 UTC |
e22c86c | lopez | 28 August 2023, 18:41:51 UTC | fix class value | 28 August 2023, 18:41:51 UTC |
300cf60 | Luca Foppiano | 25 August 2023, 02:56:18 UTC | typo | 25 August 2023, 02:56:18 UTC |
d98a71a | Luca Foppiano | 25 August 2023, 02:50:37 UTC | miss param | 25 August 2023, 02:50:37 UTC |
1910551 | Luca Foppiano | 25 August 2023, 02:45:37 UTC | read true/false value correctly | 25 August 2023, 02:45:37 UTC |
963e57a | Luca Foppiano | 25 August 2023, 02:07:26 UTC | add --early-stop true/false and decouple from max_epoch | 25 August 2023, 02:07:26 UTC |
226622a | Luca Foppiano | 24 August 2023, 05:02:57 UTC | extend multi-gpu parameter to other scripts | 24 August 2023, 05:02:57 UTC |
78f5ac6 | Luca Foppiano | 24 August 2023, 04:42:19 UTC | add parameter --max-epoch to disable early_stop and override the default values | 24 August 2023, 04:42:19 UTC |
8326458 | Luca Foppiano | 22 August 2023, 05:35:19 UTC | extend the multi-gpu to fold cross-validation | 22 August 2023, 05:35:19 UTC |
54b115d | Luca Foppiano | 22 August 2023, 02:43:03 UTC | cleanup | 22 August 2023, 02:43:03 UTC |
7afddaa | Luca Foppiano | 22 August 2023, 01:57:50 UTC | quick and dirty multi-gpu support | 22 August 2023, 02:16:04 UTC |
2f8976c | Patrice Lopez | 16 July 2023, 11:18:05 UTC | Merge pull request #161 from kermitt2/features/learning_rate_param add learning rate visualisation and manual parameter | 16 July 2023, 11:18:05 UTC |
5b3ea93 | Luca Foppiano | 16 July 2023, 01:24:17 UTC | fix default learning rate for RNN architectures | 16 July 2023, 01:24:17 UTC |
bc19e28 | Patrice Lopez | 12 July 2023, 12:15:39 UTC | missing crf output update (thanks Luca) | 12 July 2023, 12:15:39 UTC |
10503b0 | Luca Foppiano | 08 July 2023, 10:28:01 UTC | Merge branch 'master' into features/learning_rate_param | 08 July 2023, 10:28:01 UTC |
30e87a6 | Patrice Lopez | 07 July 2023, 09:12:37 UTC | Merge branch 'master' of github.com:kermitt2/delft | 07 July 2023, 09:12:37 UTC |
8ca27d3 | Patrice Lopez | 07 July 2023, 09:12:22 UTC | fix last example reader with features | 07 July 2023, 09:12:22 UTC |
64fa67c | Luca Foppiano | 15 June 2023, 02:10:11 UTC | cosmetics | 15 June 2023, 02:10:11 UTC |
edd1898 | Luca Foppiano | 15 June 2023, 01:43:11 UTC | Log learning rate at each epoque | 15 June 2023, 01:57:03 UTC |
7679402 | Luca Foppiano | 14 June 2023, 23:57:36 UTC | Merge branch 'master' into features/learning_rate_param | 14 June 2023, 23:57:36 UTC |
a405486 | Luca Foppiano | 14 June 2023, 23:34:03 UTC | Remove default value from learning rate in TrainingConfig and set different default for transformers-based and non-transformers-based models | 14 June 2023, 23:37:04 UTC |
42207e2 | Luca Foppiano | 14 June 2023, 23:32:39 UTC | Fix typo | 14 June 2023, 23:32:39 UTC |
22c2ca5 | Luca Foppiano | 14 June 2023, 23:28:45 UTC | Remove hardcoded learning rates | 14 June 2023, 23:28:45 UTC |
185428b | Luca Foppiano | 14 June 2023, 23:25:42 UTC | Fix default learning rate from applications scripts | 14 June 2023, 23:25:42 UTC |
9cd81a1 | lopez | 13 June 2023, 15:05:04 UTC | more consistent attribute name in dataset | 13 June 2023, 15:05:04 UTC |
33d6436 | lfoppiano | 12 May 2023, 23:38:30 UTC | print correct learning_rate and improve default value based on architecture | 12 May 2023, 23:38:30 UTC |
07e291b | Luca Foppiano | 08 May 2023, 08:33:37 UTC | add learning rate as parameter | 08 May 2023, 08:33:37 UTC |
5305ab9 | Luca Foppiano | 16 March 2023, 23:13:12 UTC | print error when download fails | 16 March 2023, 23:13:12 UTC |
a50fd1a | Patrice Lopez | 16 March 2023, 14:58:17 UTC | Merge pull request #157 from kermitt2/download_elmo Download ELMo embeddings is not present | 16 March 2023, 14:58:17 UTC |
efbfa16 | Patrice Lopez | 10 March 2023, 20:18:47 UTC | download ELMo embeddings is not present | 10 March 2023, 20:18:47 UTC |
c1001d4 | Patrice Lopez | 10 March 2023, 20:16:45 UTC | print patience parameter too when used | 10 March 2023, 20:16:45 UTC |
7302d4a | Patrice Lopez | 09 March 2023, 09:48:41 UTC | Merge pull request #148 from kermitt2/feature/add-patience-as-parameter Add parameters (patience, max_sequence_length, and batch_size) for sequence labelling command line scripts | 09 March 2023, 09:48:41 UTC |
fce22c5 | Luca Foppiano | 09 March 2023, 08:59:39 UTC | fix patience default value | 09 March 2023, 08:59:39 UTC |
3932827 | lfoppiano | 09 March 2023, 03:53:44 UTC | fixes for review | 09 March 2023, 03:53:44 UTC |
42cf2d8 | Patrice Lopez | 12 February 2023, 09:41:24 UTC | update dependency versions in doc | 12 February 2023, 09:41:24 UTC |
9bd493f | Patrice Lopez | 12 February 2023, 08:43:35 UTC | update to version 0.3.3 | 12 February 2023, 08:43:35 UTC |
3ee3775 | Patrice Lopez | 26 January 2023, 08:31:11 UTC | Merge pull request #154 from kermitt2/bpe-realignment Fix issue for alignment of pre-tokenized input with BPE, see #150 | 26 January 2023, 08:31:11 UTC |
6a9cfb2 | Luca Foppiano | 23 January 2023, 00:36:48 UTC | pass use_ELMO parameter to eval | 23 January 2023, 00:36:48 UTC |
73d5444 | Patrice Lopez | 19 January 2023, 16:01:39 UTC | removing non automodel luke | 19 January 2023, 16:01:39 UTC |
b5f4591 | Patrice Lopez | 19 January 2023, 16:00:17 UTC | cleaning | 19 January 2023, 16:00:17 UTC |
400181b | Patrice Lopez | 19 January 2023, 08:12:38 UTC | cleaning | 19 January 2023, 08:12:38 UTC |
e9d0399 | Patrice Lopez | 19 January 2023, 08:06:41 UTC | Merge branch 'master' into bpe-realignment | 19 January 2023, 08:06:41 UTC |
88f084d | Patrice Lopez | 19 January 2023, 07:19:12 UTC | Merge pull request #156 from kermitt2/fix-144 fix #144 for defaulting case | 19 January 2023, 07:19:12 UTC |
73dc549 | Luca Foppiano | 19 January 2023, 02:10:41 UTC | Testing the lowercase name, just in case | 19 January 2023, 02:10:41 UTC |
47fe8a7 | Patrice Lopez | 18 January 2023, 15:34:03 UTC | fix #144 for defaulting case | 18 January 2023, 15:34:03 UTC |
5aea590 | Patrice Lopez | 18 January 2023, 15:31:31 UTC | fix for some sentencepiece tokenizer | 18 January 2023, 15:31:31 UTC |
6806d12 | Patrice Lopez | 17 January 2023, 23:08:41 UTC | fix filtering for some roberta variants | 17 January 2023, 23:08:41 UTC |
befd674 | Patrice Lopez | 17 January 2023, 12:35:14 UTC | fix issue for alignment of pre-tokenized input with BPE, see #150 | 17 January 2023, 12:35:14 UTC |
2641dd9 | dependabot[bot] | 13 January 2023, 19:14:37 UTC | Bump torch from 1.10.1 to 1.13.1 Bumps [torch](https://github.com/pytorch/pytorch) from 1.10.1 to 1.13.1. - [Release notes](https://github.com/pytorch/pytorch/releases) - [Changelog](https://github.com/pytorch/pytorch/blob/master/RELEASE.md) - [Commits](https://github.com/pytorch/pytorch/compare/v1.10.1...v1.13.1) --- updated-dependencies: - dependency-name: torch dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> | 13 January 2023, 19:14:37 UTC |
e38e2df | Patrice Lopez | 13 January 2023, 19:13:51 UTC | make requirements.txt consistent with setup.py | 13 January 2023, 19:13:51 UTC |
070d2ba | Luca Foppiano | 06 January 2023, 01:59:19 UTC | do not cancel build if anything else is in progress | 06 January 2023, 01:59:19 UTC |
82ff774 | lopez | 02 January 2023, 17:36:31 UTC | cleaning | 02 January 2023, 17:36:31 UTC |
c054783 | lopez | 02 January 2023, 14:54:48 UTC | fix table | 02 January 2023, 14:54:48 UTC |
59a41b0 | Patrice Lopez | 02 January 2023, 14:49:11 UTC | some benchmarking for models with and without features | 02 January 2023, 14:49:11 UTC |
149cc40 | Patrice Lopez | 24 December 2022, 11:27:19 UTC | add a BERT_FEATURES architecture, no CRF | 24 December 2022, 11:27:19 UTC |
1a894a6 | Patrice Lopez | 22 December 2022, 09:30:04 UTC | Merge branch 'master' of github.com:kermitt2/delft | 22 December 2022, 09:30:04 UTC |
afe611b | Patrice Lopez | 22 December 2022, 09:29:44 UTC | add BERT_ChainCRF_FEATURES variant | 22 December 2022, 09:29:44 UTC |
76057df | Luca Foppiano | 19 December 2022, 08:04:19 UTC | move import for plotting model | 19 December 2022, 08:04:19 UTC |
cec63f9 | Luca Foppiano | 19 December 2022, 07:55:28 UTC | add patience and max_sequence_length and batch size in command line | 19 December 2022, 07:55:28 UTC |
389eb3d | Patrice Lopez | 10 December 2022, 12:32:57 UTC | update main dependencies | 10 December 2022, 12:32:57 UTC |
ea328ed | Patrice Lopez | 09 December 2022, 19:29:55 UTC | refer to PyPI | 09 December 2022, 19:29:55 UTC |
063da78 | Luca Foppiano | 08 December 2022, 01:49:22 UTC | add reader of crf files without labels | 08 December 2022, 01:54:48 UTC |
e8d3ab6 | Luca Foppiano | 08 December 2022, 01:49:03 UTC | cosmetics | 08 December 2022, 01:54:48 UTC |
9cbcf90 | Patrice Lopez | 25 November 2022, 16:47:22 UTC | Merge pull request #147 from kermitt2/incremental-training Incremental training | 25 November 2022, 16:47:22 UTC |
7ca4ca9 | Patrice Lopez | 24 November 2022, 20:35:58 UTC | extend vocab and tags in case of incremental training | 24 November 2022, 20:35:58 UTC |
ceebe28 | Patrice Lopez | 24 November 2022, 13:19:36 UTC | try to fix github action tests | 24 November 2022, 13:19:36 UTC |
345107c | Patrice Lopez | 24 November 2022, 12:45:04 UTC | fix typo :/ | 24 November 2022, 12:45:04 UTC |
5de037d | Patrice Lopez | 24 November 2022, 10:14:19 UTC | allow to set model path for incremental training | 24 November 2022, 10:14:19 UTC |