 Kaushik Ram Sadagopan
|
a11376477b
Add use_text_decoder parameter to UnitYConfig. (#88)
|
vor 2 Jahren |
 Can Balioglu
|
05419775be
Introduce Prosody encoder (#87)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
bfcdf3ba4f
Skips loading the vocoder for X2T eval. (#84)
|
vor 2 Jahren |
|
Can Balioglu
|
9daf4692ef
Revise project structure (#82)
|
vor 2 Jahren |
 spopuri
|
a1646750f1
Add evaluation pipeline to compute all metrics (#79)
|
vor 2 Jahren |
|
Can Balioglu
|
74364ff4db
Derive UnitYT2UModel from EncoderDecoderModel (#77)
|
vor 2 Jahren |
|
Can Balioglu
|
e83a4de3af
Start using vocab_info instead of pad_idx (#76)
|
vor 2 Jahren |
|
Can Balioglu
|
04d9a49462
Update LogitsProcessor to StepProcessor (#74)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
45974cd2f4
Disintegrate UnitYNART2UModel's decoder and decoder_layer from TransformerDecoder, TransformerDecoderLayer. (#71)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
4d679cb67e
Separate T2U and NAR T2U builders for UnitY T2U model. (#70)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
c60f92c86f
Rename 'commercial' to 'v2' and condition duration prediction on T2U model type for vocoder. (#72)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
f4dffda0f8
Translator integration tests and unity unittests. (#63)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
3b34f6defa
Don't mandate audio_root_dir for text input eval. (#69)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
eba940ace3
m4t evaluation script (#57)
|
vor 2 Jahren |
|
Can Balioglu
|
bc32323fa2
Hook (#66)
|
vor 2 Jahren |
|
Can Balioglu
|
dc9dd18cae
Introduces create_raw_encoder in CharTokenizer (#65)
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
4314c30490
Run black on the entire repo, fix mypy issues in translator.py, fix API bug in unity/generator.py (#62)
|
vor 2 Jahren |
|
Can Balioglu
|
b91e141a16
Rename pad_idx to pad_value in Collater (#60)
|
vor 2 Jahren |
 Ruslan Mavlyutov
|
80fd023318
Merge pull request #58 from fairinternal/ruslan_m4t_training_scripts_follow_ups
|
vor 2 Jahren |
|
mavlyutov
|
5c35042f34
M4T training scripts follow-ups
|
vor 2 Jahren |
|
Ruslan Mavlyutov
|
2477f1d753
Merge pull request #56 from fairinternal/ruslan_update_training_recipes
|
vor 2 Jahren |
|
mavlyutov
|
c4e2953141
Update training recipees for asr and M4T large; Fix GPU ooms with M4T large training
|
vor 2 Jahren |
|
Ruslan Mavlyutov
|
9ac6068700
Merge pull request #45 from fairinternal/ruslan_m4t_training
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
f45bd3a4b9
Switching out seamlessM4T_v2_large to the multitask model with the text encoder. (#55)
|
vor 2 Jahren |
|
mavlyutov
|
d8c45cc9e7
Adjust for latest changes in fairseq2n
|
vor 2 Jahren |
|
mavlyutov
|
efe88afa2e
M4T training scripts and recipes
|
vor 2 Jahren |
|
Kaushik Ram Sadagopan
|
d8521ec060
Padding mask bug fix, it is boolean and not float, UnitY generator. (#54)
|
vor 2 Jahren |
|
Can Balioglu
|
1bb19a6aa1
Model dim (#53)
|
vor 2 Jahren |
|
Can Balioglu
|
d314c9c4da
Statickv (#52)
|
vor 2 Jahren |
 Pierre Andrews
|
8a1808944c
Merge pull request #51 from fairinternal/commercial_vocoder_asset
|
vor 2 Jahren |
