NeMo/nemo
Samuel Kriman b7a175b7b9
Self-supervised pre-training for speech models (#3139)
* self-supervised training

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* test

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* remove imports

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* fix

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* sort imports

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* fix audio_to_text

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* manifest handle no text

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* loss init

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* style

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* remove tokenizer from config

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* config changes

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* remove hydra import

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* always spec augment

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* fixes

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* copyright

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* fix cosine sim

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* fix cosine sim

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* fix cosine sim

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* changes based on comments

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* changes based on comments

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* configs

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* name fix

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* ci config changes

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* renamed to num_negatives

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* minor changes

Signed-off-by: sam1373 <samuelkriman@gmail.com>

* name changes, type annotations

Signed-off-by: sam1373 <samuelkriman@gmail.com>

Co-authored-by: Yang Zhang <yzhang123@users.noreply.github.com>
2021-11-10 15:33:11 -08:00
..
collections Self-supervised pre-training for speech models (#3139) 2021-11-10 15:33:11 -08:00
core Self-supervised pre-training for speech models (#3139) 2021-11-10 15:33:11 -08:00
utils Merge r1.5.0 bugfixes and doc updates to main (#3133) 2021-11-04 10:26:58 -06:00
__init__.py Text denormalization (#1797) 2021-03-31 13:31:19 -07:00
constants.py Add torchaudio support to Dockerfile (#1216) 2020-09-24 23:47:06 -07:00
package_info.py Merge r1.5.0 bugfixes and doc updates to main (#3133) 2021-11-04 10:26:58 -06:00
README.md clean-plate commit 2020-06-19 22:26:00 -07:00

NeMo (Neural Modules) is a toolkit for creating AI applications built around neural modules, conceptual blocks of neural networks that take typed inputs and produce typed outputs.

NeMo Core provides common APIs all modules and models have to implement.

NeMo Collections

  • ASR - collection of modules and models for building speech recognition networks
  • TTS - collection of modules and models for building speech synthesis networks
  • NLP - collection of modules and models for building NLP networks