Commits · jesseplusplus/easy-translate

Implement SeamlessM4T

9dcafee

Iker commited on Nov 30, 2023

Add support for SMALL100 and update examples

911ab5a

Iker commited on Jun 23, 2023

Add warning to prevent large batch sizes when device map is set to auto

fbc4e83

Iker commited on Jun 23, 2023

8 bits / 4 bits quantization, load any model, promoting.

d54a92e

Iker commited on Jun 18, 2023

Add --repetition-penalty flag

390a692

Iker commited on May 27, 2023

Add --keep_tokenization_spaces argument to control the space decoding

2a897d7

Iker commited on May 26, 2023

feat: Add --keep_special_tokens argument to control special token decoding

086ac2b

ruanchaves commited on Apr 18, 2023

Fix output if translator predicts new line in Windows format

f016c88

Iker commited on Mar 19, 2023

Fix output if translator predicts new line

3d26fd2

Iker commited on Mar 19, 2023

Current output if translation predicts new line

0f4f627

Iker commited on Mar 18, 2023

Test if the source language is supported by the tokenizer

998d5ca

Iker commited on Nov 9, 2022

Disable dataloader multiprocessing

ee0f30d

Iker commited on Oct 11, 2022

Fix bug with file loading

0af8cb0

Iker commited on Oct 11, 2022

Fix find_executable_batch_size deprecation

d3132fd

Iker commited on Sep 1, 2022

Manually calculate dataloader len

5bd7f14

Iker commited on Sep 1, 2022

Count lines instead of using dataloader.dataset

011cb1f

Iker commited on Sep 1, 2022

Set num_workers to 1 manually

9a2fa00

Iker commited on Sep 1, 2022

Test params

1cb8a41

Iker commited on Sep 1, 2022

Test params

7f820ab

Iker commited on Sep 1, 2022

Test params

5829755

Iker commited on Sep 1, 2022

Test params

880922b

Iker commited on Sep 1, 2022

Fix multi-GPU batch split bug

01e13d9

Iker commited on Sep 1, 2022

Support for NLLB and Sampling

5401d1a

Iker commited on Sep 1, 2022

Fix tqdm

a4abd95

Iker commited on Jul 1, 2022

Implement num_return_sequences parameter.

95d0c3a

Iker commited on Jul 1, 2022

Fix bug

56e0241

Iker commited on Jul 1, 2022

Fix bug

ea68187

Iker commited on May 29, 2022

Fix bug

57ffa19

Iker commited on May 29, 2022

Fix: Remove duplicates when using more than 1 device

4db19c9

Iker commited on May 19, 2022

Print in standard format

ed535ee

Iker commited on May 13, 2022

Print only device type

774b679

Iker commited on May 5, 2022

[Fix Bug] Max length param was ignored

9f0798a

Iker commited on May 5, 2022

Print run info only in main process

02502dd

Iker commited on May 5, 2022

Print run info

f85b100

Iker commited on May 5, 2022

Print model name

122365e

Iker commited on May 5, 2022

Batch size description

a4d80df

Iker commited on May 4, 2022

Fix error on multi-gpu eval

6cc3f7d

Iker commited on May 4, 2022

Implement evaluation

62b1ca5

Iker commited on May 4, 2022

M2M100 with transformers and accelerate

b873cb9

Iker commited on May 4, 2022

Fix bugs

749ff6d

Iker commited on Apr 26, 2022

Initial commit

1e19e28

Iker commited on Apr 26, 2022

Commit History

Implement SeamlessM4T 9dcafee

Add support for SMALL100 and update examples 911ab5a

Add warning to prevent large batch sizes when device map is set to auto fbc4e83

8 bits / 4 bits quantization, load any model, promoting. d54a92e

Add --repetition-penalty flag 390a692

Add --keep_tokenization_spaces argument to control the space decoding 2a897d7

feat: Add --keep_special_tokens argument to control special token decoding 086ac2b

Fix output if translator predicts new line in Windows format f016c88

Fix output if translator predicts new line 3d26fd2

Current output if translation predicts new line 0f4f627

Test if the source language is supported by the tokenizer 998d5ca

Disable dataloader multiprocessing ee0f30d

Fix bug with file loading 0af8cb0

Fix find_executable_batch_size deprecation d3132fd

Manually calculate dataloader len 5bd7f14

Count lines instead of using dataloader.dataset 011cb1f

Set num_workers to 1 manually 9a2fa00

Test params 1cb8a41

Test params 7f820ab

Test params 5829755

Test params 880922b

Fix multi-GPU batch split bug 01e13d9

Support for NLLB and Sampling 5401d1a

Fix tqdm a4abd95

Implement num_return_sequences parameter. 95d0c3a

Fix bug 56e0241

Fix bug ea68187

Fix bug 57ffa19

Fix: Remove duplicates when using more than 1 device 4db19c9

Print in standard format ed535ee

Print only device type 774b679

[Fix Bug] Max length param was ignored 9f0798a

Print run info only in main process 02502dd

Print run info f85b100

Print model name 122365e

Batch size description a4d80df

Fix error on multi-gpu eval 6cc3f7d

Implement evaluation 62b1ca5

M2M100 with transformers and accelerate b873cb9

Fix bugs 749ff6d

Initial commit 1e19e28

Implement SeamlessM4T

9dcafee

Add support for SMALL100 and update examples

911ab5a

Add warning to prevent large batch sizes when device map is set to auto

fbc4e83

8 bits / 4 bits quantization, load any model, promoting.

d54a92e

Add --repetition-penalty flag

390a692

Add --keep_tokenization_spaces argument to control the space decoding

2a897d7

feat: Add --keep_special_tokens argument to control special token decoding

086ac2b

Fix output if translator predicts new line in Windows format

f016c88

Fix output if translator predicts new line

3d26fd2

Current output if translation predicts new line

0f4f627

Test if the source language is supported by the tokenizer

998d5ca

Disable dataloader multiprocessing

ee0f30d

Fix bug with file loading

0af8cb0

Fix find_executable_batch_size deprecation

d3132fd

Manually calculate dataloader len

5bd7f14

Count lines instead of using dataloader.dataset

011cb1f

Set num_workers to 1 manually

9a2fa00

Test params

1cb8a41

Test params

7f820ab

Test params

5829755

Test params

880922b

Fix multi-GPU batch split bug

01e13d9

Support for NLLB and Sampling

5401d1a

Fix tqdm

a4abd95

Implement num_return_sequences parameter.

95d0c3a

Fix bug

56e0241

Fix bug

ea68187

Fix bug

57ffa19

Fix: Remove duplicates when using more than 1 device

4db19c9

Print in standard format

ed535ee

Print only device type

774b679

[Fix Bug] Max length param was ignored

9f0798a

Print run info only in main process

02502dd

Print run info

f85b100

Print model name

122365e

Batch size description

a4d80df

Fix error on multi-gpu eval

6cc3f7d

Implement evaluation

62b1ca5

M2M100 with transformers and accelerate

b873cb9

Fix bugs

749ff6d

Initial commit

1e19e28