Commit Graph

934 Commits

Author SHA1 Message Date
Tarun Menta
d3aecc0977
Pick correct dtype on T4 GPUs
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-23 17:22:34 -04:00
Tarun Menta
eb179cc543
Update layout batch sizes 2025-09-23 17:09:09 -04:00
Tarun Menta
9bee27c4e0
Fix tests 2025-09-23 16:57:27 -04:00
Tarun Menta
1d09025ea3
Bump foundation checkpoint 2025-09-23 16:39:10 -04:00
Tarun Menta
4d7be669ab
Models moved to S3 2025-09-23 16:33:05 -04:00
Tarun Menta
5811d072f4
Separate models for layout and OCR 2025-09-23 15:39:44 -04:00
Tarun Menta
c1719e982c
Update loaders with dtype instead of torch_dtype -- transformers
`torch_dtype` was deprecated, will silently fail
2025-09-22 12:49:51 -04:00
Tarun Menta
42e016f49c
Update tqdm desc string based on founation model mode
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-19 12:51:01 -04:00
Tarun Menta
9ab25b3ede
Merge branch 'dev' into layout-release 2025-09-19 12:47:06 -04:00
Tarun Menta
20f9179503
Pin sliding window for layout 2025-09-18 15:51:32 -04:00
Tarun Menta
74e790ccd8
Tokenizer fix 2025-09-17 16:06:15 -04:00
Tarun Menta
9673ec6abe
Merge in recognition predictor changes from dev 2025-09-16 21:27:53 -04:00
Tarun Menta
2d5dd9b86f
Move back to old table_rec model for now 2025-09-16 21:18:37 -04:00
Vik Paruchuri
a7ffa7ee51
Merge pull request #457 from datalab-to/dev
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
Move flash attention funcs
2025-09-08 12:38:58 -04:00
Vik Paruchuri
5234bc09c2 Move flash attention funcs
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-08 12:21:52 -04:00
Vik Paruchuri
0a4068b40a Iterate on bbox head
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-08 11:40:40 -04:00
Vik Paruchuri
a7b133ec77
Merge pull request #456 from datalab-to/dev
Enable setting attention method
2025-09-08 11:32:52 -04:00
Vik Paruchuri
ebf5ec72e1
Merge pull request #455 from datalab-to/vik/no_autoset
Get rid of attention method checks
2025-09-08 11:28:02 -04:00
Vik Paruchuri
a4c3c85279 Bump transformers 2025-09-08 09:58:46 -04:00
Vik Paruchuri
3c9dfe48bf Fix tests 2025-09-08 09:39:35 -04:00
Vik Paruchuri
605339d68b Better heuristics if implementation is None 2025-09-08 09:24:40 -04:00
Vik Paruchuri
4ed13823fc Get rid of attention method checks 2025-09-08 09:04:19 -04:00
Vik Paruchuri
b0b18176ed
Merge pull request #454 from datalab-to/dev
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
Foundation predictor init
2025-09-07 21:40:36 -04:00
Vik Paruchuri
cf6dbde72c Foundation predictor init
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-07 21:39:20 -04:00
Vik Paruchuri
2813924867
Merge pull request #453 from datalab-to/dev
Bump version
2025-09-07 21:17:12 -04:00
Vik Paruchuri
35d2734412 Bump version 2025-09-07 21:12:42 -04:00
Vik Paruchuri
baca4df27c
Merge pull request #452 from datalab-to/dev
Dev
2025-09-07 21:06:37 -04:00
Vik Paruchuri
43e74bcd95
Merge pull request #451 from datalab-to/sdpa-fix
Foundation Model Performance Improvements
2025-09-07 20:52:46 -04:00
Vik Paruchuri
7da76d18cc Add attention implementation 2025-09-07 20:48:04 -04:00
Vik Paruchuri
3b1e8dc51d Add bbox head 2025-09-07 20:41:56 -04:00
Tarun Menta
b96ac10282
Sort by total image pixels, since we are in block mode now 2025-09-07 10:31:50 -04:00
Tarun Menta
179848b037
Fix cache positions for SDPA 2025-09-06 19:30:43 -04:00
Vik Paruchuri
2db254c7e8
Merge pull request #449 from datalab-to/master
Backport
2025-09-05 16:29:13 -04:00
Zach Nussbaum
dca1f48dd2
bump version 2025-09-05 19:58:31 +00:00
Zach Nussbaum
7b600a5886
bump version 2025-09-05 19:15:53 +00:00
Zach Nussbaum
13beff2723
Merge pull request #448 from datalab-to/dev
Dev
2025-09-05 15:13:33 -04:00
Zach Nussbaum
273040d8fc
Merge pull request #446 from datalab-to/multi-token
feat: multi-token decoding
2025-09-05 15:05:59 -04:00
Zach Nussbaum
8821050961
fix: update model
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-05 18:49:04 +00:00
Zach Nussbaum
b8a464c881
fix: opencv version fix 2025-09-05 18:32:51 +00:00
Ashish Uppala
1a32fa6cd1
Update README 2025-09-05 11:31:46 -04:00
github-actions[bot]
950559b5fd
@u-ashish has signed the CLA in datalab-to/surya#447 2025-09-05 15:16:58 +00:00
u-ashish
d9121e17bc Update README 2025-09-05 11:14:17 -04:00
Zach Nussbaum
691030445f
refactor: move min confidence to settings 2025-09-04 00:42:26 +00:00
Zach Nussbaum
9282294356
fix: don't do position rolling bc it's handled already 2025-09-03 21:37:26 +00:00
Zach Nussbaum
0ef78da9b4
Merge branch 'dev' into multi-token 2025-09-03 21:09:05 +00:00
Zach Nussbaum
2c172f2fb2
feat: not working conditional decode 2025-09-03 21:05:43 +00:00
github-actions[bot]
0410f44270
@davidxifeng has signed the CLA in datalab-to/surya#445
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-03 14:52:28 +00:00
Vik Paruchuri
4acbc8e86d
Merge pull request #444 from datalab-to/dev
Remove unused import
2025-09-02 14:25:01 -04:00
Vik Paruchuri
5715935c48 Remove unused import
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
2025-09-02 13:56:46 -04:00
Vik Paruchuri
a8a021490e
Merge pull request #442 from datalab-to/new-tokenizer
Some checks failed
Integration test / build (push) Has been cancelled
Unit tests / build (t4_gpu) (push) Has been cancelled
Unit tests / build (ubuntu-latest) (push) Has been cancelled
Unit tests / build (windows-latest) (push) Has been cancelled
Test CLI scripts / build (push) Has been cancelled
feat: new unified tokenizer
2025-08-30 07:24:25 +02:00