Skip to content

Commit cf83c5f

Browse files
XuesongYangericharperredoctopusVahidooXZhilin123
authored
merge r1.11 to main (#4920)
* update branch Signed-off-by: ericharper <[email protected]> * update package info and dockerfile Signed-off-by: ericharper <[email protected]> * [TTS] bugfix for missing configs. (#4725) Signed-off-by: Xuesong Yang <[email protected]> * Fix pynini install in TTS tutorials (#4729) Signed-off-by: Jocelyn Huang <[email protected]> Signed-off-by: Jocelyn Huang <[email protected]> * [TTS] updated config with a German IPA phoneme tokenizer (#4756) * [TTS] added a German IPA phoneme tokenizer * [TTS][ASR] enabled customized arguments for trimming the leading and trailing silence. * [TTS] disabled spline interpolation for beta-binomial distribution. Let it generate align prior and save to disks. Use a new phoneme tokenizer. * [TTS] use consistent spline interpolation with fastpitch checkpoint when generating mel-spectrograms for hifigan finetune. Signed-off-by: Xuesong Yang <[email protected]> * Update r1.11 to new heteronyms list (#4745) * Update configs to new heteronyms list * Remove old heteronyms list, add alt 'merchandise' pron to CMUdict * Update remaining references to old heteronyms list Signed-off-by: Jocelyn Huang <[email protected]> Co-authored-by: Xuesong Yang <[email protected]> * Fix tutorial formatting (#4778) Signed-off-by: Jocelyn Huang <[email protected]> * update branch and typos (#4788) Signed-off-by: ericharper <[email protected]> Signed-off-by: ericharper <[email protected]> * Adding support for models trained with full context for cache-aware streaming. (#4687) * added support for models trained with full context. Signed-off-by: Vahid <[email protected]> * fixed style. Signed-off-by: Vahid <[email protected]> * dropped seq_range Signed-off-by: Vahid <[email protected]> * fixed indexing in caching methods. Signed-off-by: Vahid <[email protected]> * fixed code style. Signed-off-by: Vahid <[email protected]> * fixed code style. Signed-off-by: Vahid <[email protected]> * updated docs. Signed-off-by: Vahid <[email protected]> * addressed comments. Signed-off-by: Vahid <[email protected]> * fixed code style. Signed-off-by: Vahid <[email protected]> * fixed code style. Signed-off-by: Vahid <[email protected]> * fixed code style. Signed-off-by: Vahid <[email protected]> * change frame-wise to cache-aware. Signed-off-by: Vahid <[email protected]> * change frame-wise to cache-aware. Signed-off-by: Vahid <[email protected]> * change frame-wise to cache-aware. Signed-off-by: Vahid <[email protected]> * fixed code style. Signed-off-by: Vahid <[email protected]> Signed-off-by: Vahid <[email protected]> * Update megatron encoder decoder model to support py37 for colab (#4791) * [ASR] Add pretrained ASR models for Croatian (#4682) * [ASR] Add pretrained ASR models for Croatian Signed-off-by: Ante Jukić <[email protected]> * Fix style for import Signed-off-by: Ante Jukić <[email protected]> Signed-off-by: Ante Jukić <[email protected]> Co-authored-by: Ante Jukić <[email protected]> Co-authored-by: Nithin Rao <[email protected]> Co-authored-by: Eric Harper <[email protected]> Co-authored-by: Somshubra Majumdar <[email protected]> * added/fixed export for Megatron models (#4712) * added/fixed export for Megatron models Signed-off-by: David Mosallanezhad <[email protected]> * fixed style Signed-off-by: David Mosallanezhad <[email protected]> * fixed FusedScaleMaskSoftmax in BioMegatron Signed-off-by: David Mosallanezhad <[email protected]> * included comments Signed-off-by: David Mosallanezhad <[email protected]> Signed-off-by: David Mosallanezhad <[email protected]> Co-authored-by: David Mosallanezhad <[email protected]> Co-authored-by: Eric Harper <[email protected]> * update branch for qa notebook Signed-off-by: ericharper <[email protected]> * Fix initializing weights from ptl ckpt with exclude (#4807) Signed-off-by: sam1373 <[email protected]> Signed-off-by: sam1373 <[email protected]> * Fix index error from addition of voiced_mask and p_voiced (#4811) Signed-off-by: Jocelyn Huang <[email protected]> Signed-off-by: Jocelyn Huang <[email protected]> * T5 prompt learning fixes (#4771) * RPE, hidden size and config fixes Signed-off-by: MaximumEntropy <[email protected]> * Update to reflect new config names Signed-off-by: MaximumEntropy <[email protected]> * Sentencepiece fixes Signed-off-by: MaximumEntropy <[email protected]> * Style Signed-off-by: MaximumEntropy <[email protected]> * Fix finetuning Signed-off-by: MaximumEntropy <[email protected]> * Add encoder seq len to gpt Signed-off-by: MaximumEntropy <[email protected]> * Style Signed-off-by: MaximumEntropy <[email protected]> * Add finetune eval script Signed-off-by: MaximumEntropy <[email protected]> * Fix name Signed-off-by: MaximumEntropy <[email protected]> * Update Jenkinsfile Signed-off-by: MaximumEntropy <[email protected]> * Update config Signed-off-by: MaximumEntropy <[email protected]> * Fix CI test Signed-off-by: MaximumEntropy <[email protected]> * Update check Signed-off-by: MaximumEntropy <[email protected]> * Style Signed-off-by: MaximumEntropy <[email protected]> * Backward compat Signed-off-by: MaximumEntropy <[email protected]> * Update CI test Signed-off-by: MaximumEntropy <[email protected]> * Split rank for Enc-Dec models Signed-off-by: MaximumEntropy <[email protected]> * Address comments Signed-off-by: MaximumEntropy <[email protected]> * Style Signed-off-by: MaximumEntropy <[email protected]> Signed-off-by: MaximumEntropy <[email protected]> Co-authored-by: Virginia Adams <[email protected]> * G2P docs (#4841) * g2p docs added Signed-off-by: ekmb <[email protected]> * fix references Signed-off-by: ekmb <[email protected]> * address review feedback Signed-off-by: ekmb <[email protected]> Signed-off-by: ekmb <[email protected]> * Fix providing glue in seq2seq eval (#4843) * Fix providing glue in seq2seq eval Signed-off-by: MaximumEntropy <[email protected]> * Fix Signed-off-by: MaximumEntropy <[email protected]> * Style Signed-off-by: MaximumEntropy <[email protected]> Signed-off-by: MaximumEntropy <[email protected]> * Updated inference code and squad scripts (#4835) * Updated inference code and squad scripts Signed-off-by: Virginia Adams <[email protected]> * Reverted GPT & T5 inference files back to use NLPDDPlugin Signed-off-by: Virginia Adams <[email protected]> * Overwrite frozen LM to use fused adam Signed-off-by: Virginia Adams <[email protected]> * Added padded vocab size Signed-off-by: Virginia Adams <[email protected]> * Fixed val check interval value Signed-off-by: Virginia Adams <[email protected]> * Python format fix Signed-off-by: Virginia Adams <[email protected]> * Make t5 prompt learning preds write to file Signed-off-by: Virginia Adams <[email protected]> * Added back dp=1 check Signed-off-by: Virginia Adams <[email protected]> Signed-off-by: Virginia Adams <[email protected]> Co-authored-by: Sandeep Subramanian <[email protected]> * Set the number of workers to 0 for validation and test sets in all enc-dec models (#4790) * Set workers to 0 for validation and test Signed-off-by: MaximumEntropy <[email protected]> * Revert pin memory Signed-off-by: MaximumEntropy <[email protected]> * Style Signed-off-by: MaximumEntropy <[email protected]> Signed-off-by: MaximumEntropy <[email protected]> Co-authored-by: Sean Naren <[email protected]> * Fix Megatron NMT consumed samples and ckpt_to_nemo split rank (#4884) * Fix nmt and ckpt_to_nemo Signed-off-by: MaximumEntropy <[email protected]> * Style Signed-off-by: MaximumEntropy <[email protected]> Signed-off-by: MaximumEntropy <[email protected]> * added utf8 encoding (#4892) Signed-off-by: Virginia Adams <[email protected]> Signed-off-by: Virginia Adams <[email protected]> * update readme with apex commit Signed-off-by: ericharper <[email protected]> * Add support for Apex distributed Adam optimizer with GPT-3 (#4487) * Add support for Apex distributed Adam optimizer with GPT-3 Signed-off-by: Tim Moon <[email protected]> * Fix bug in grad clipping with dist Adam Grad norm was computed over all params, not respecting model parallelism. Signed-off-by: Tim Moon <[email protected]> * Fix bug with DDP initialization Signed-off-by: Tim Moon <[email protected]> * Make distopt dependent on megatron_amp_o2 Signed-off-by: Tim Moon <[email protected]> * Fix code formatting Signed-off-by: Tim Moon <[email protected]> * Handle dist Adam in optimizer unit tests Signed-off-by: Tim Moon <[email protected]> Signed-off-by: Tim Moon <[email protected]> Co-authored-by: Eric Harper <[email protected]> * update readme Signed-off-by: ericharper <[email protected]> * update readme Signed-off-by: ericharper <[email protected]> * fixed styles Signed-off-by: Xuesong Yang <[email protected]> * removed unsued import. Signed-off-by: Xuesong Yang <[email protected]> * removed duplicated func defintion. Signed-off-by: Xuesong Yang <[email protected]> * replace 'r1.11.0' with 'main' in Jenkinsfile and all tutorials. Signed-off-by: Xuesong Yang <[email protected]> * fix: PRE_RELEASE = 'rc0' Signed-off-by: Xuesong Yang <[email protected]> * replace branch name to main for asr_with_adapters.ipynb. Signed-off-by: Xuesong Yang <[email protected]> * fix Fastpitch mixertts tutorial format to align with main to distingshuish diff Signed-off-by: Xuesong Yang <[email protected]> * fix: correct path for tokenizers. Signed-off-by: Xuesong Yang <[email protected]> Signed-off-by: ericharper <[email protected]> Signed-off-by: Xuesong Yang <[email protected]> Signed-off-by: Jocelyn Huang <[email protected]> Signed-off-by: Vahid <[email protected]> Signed-off-by: Ante Jukić <[email protected]> Signed-off-by: David Mosallanezhad <[email protected]> Signed-off-by: sam1373 <[email protected]> Signed-off-by: MaximumEntropy <[email protected]> Signed-off-by: ekmb <[email protected]> Signed-off-by: Virginia Adams <[email protected]> Signed-off-by: Tim Moon <[email protected]> Co-authored-by: ericharper <[email protected]> Co-authored-by: Jocelyn <[email protected]> Co-authored-by: Vahid Noroozi <[email protected]> Co-authored-by: Zhilin Wang <[email protected]> Co-authored-by: anteju <[email protected]> Co-authored-by: Ante Jukić <[email protected]> Co-authored-by: Nithin Rao <[email protected]> Co-authored-by: Somshubra Majumdar <[email protected]> Co-authored-by: David <[email protected]> Co-authored-by: David Mosallanezhad <[email protected]> Co-authored-by: Samuel Kriman <[email protected]> Co-authored-by: Sandeep Subramanian <[email protected]> Co-authored-by: Virginia Adams <[email protected]> Co-authored-by: Evelina <[email protected]> Co-authored-by: Sean Naren <[email protected]> Co-authored-by: Tim Moon <[email protected]>
1 parent 4f6aba1 commit cf83c5f

File tree

68 files changed

+1533
-359
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

68 files changed

+1533
-359
lines changed

Jenkinsfile

+12-4
Original file line numberDiff line numberDiff line change
@@ -3146,8 +3146,10 @@ assert_frame_equal(training_curve, gt_curve, rtol=1e-3, atol=1e-3)"'''
31463146
inference.add_BOS=False \
31473147
trainer.devices=2 \
31483148
tensor_model_parallel_size=2 \
3149+
pred_file_path=/home/TestData/nlp/prompt_learning/p_tuning_test_tp_preds.txt \
31493150
data_paths=['/home/TestData/nlp/prompt_learning/rte_CI_test.jsonl']"
31503151
sh "rm -rf /home/TestData/nlp/prompt_learning/p_tuning_test_tp.nemo"
3152+
sh "rm -rf /home/TestData/nlp/prompt_learning/p_tuning_test_tp_preds.txt"
31513153
}
31523154
}
31533155
stage('GPT Prompt Learning TP=1 PP=2') {
@@ -3173,8 +3175,10 @@ assert_frame_equal(training_curve, gt_curve, rtol=1e-3, atol=1e-3)"'''
31733175
inference.add_BOS=False \
31743176
trainer.devices=2 \
31753177
pipeline_model_parallel_size=2 \
3178+
pred_file_path=/home/TestData/nlp/prompt_learning/p_tuning_test_pp_preds.txt \
31763179
data_paths=['/home/TestData/nlp/prompt_learning/boolq_CI_test.jsonl']"
31773180
sh "rm -rf /home/TestData/nlp/prompt_learning/p_tuning_test_pp.nemo"
3181+
sh "rm -rf /home/TestData/nlp/prompt_learning/p_tuning_test_pp_preds.txt"
31783182
}
31793183
}
31803184
}
@@ -3433,7 +3437,7 @@ assert_frame_equal(training_curve, gt_curve, rtol=1e-3, atol=1e-3)"'''
34333437
trainer.max_steps=6 \
34343438
trainer.max_epochs=null \
34353439
model.tensor_model_parallel_size=1 \
3436-
model.pretrained_language_model_path='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m-refactor.nemo' \
3440+
model.language_model_path='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m-refactor.nemo' \
34373441
model.existing_tasks=[] \
34383442
model.new_tasks=['squad'] \
34393443
model.data.train_ds=['/home/TestData/nlp/prompt_learning/squad_CI_test.jsonl'] \
@@ -3443,11 +3447,13 @@ assert_frame_equal(training_curve, gt_curve, rtol=1e-3, atol=1e-3)"'''
34433447
sh "rm -rf /home/TestData/nlp/prompt_learning/t5_p_tuning_test"
34443448
sh "python examples/nlp/language_modeling/megatron_t5_prompt_learning_eval.py \
34453449
virtual_prompt_model_file='/home/TestData/nlp/prompt_learning/t5_p_tuning_test.nemo' \
3446-
pretrained_language_model_file='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m-refactor.nemo' \
3450+
language_model_path='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m-refactor.nemo' \
34473451
data.test_ds=['/home/TestData/nlp/prompt_learning/squad_CI_test.jsonl'] \
3452+
pred_file_path='/home/TestData/nlp/prompt_learning/t5_p_tuning_test_preds.txt' \
34483453
data.global_batch_size=4 \
34493454
data.micro_batch_size=4"
34503455
sh "rm -rf /home/TestData/nlp/prompt_learning/t5_p_tuning_test.nemo"
3456+
sh "rm -rf /home/TestData/nlp/prompt_learning/t5_p_tuning_test_preds.txt"
34513457
}
34523458
}
34533459
stage('T5 Prompt Learning TP=2 PP=1') {
@@ -3459,7 +3465,7 @@ assert_frame_equal(training_curve, gt_curve, rtol=1e-3, atol=1e-3)"'''
34593465
trainer.max_steps=6 \
34603466
trainer.max_epochs=null \
34613467
model.tensor_model_parallel_size=2 \
3462-
model.pretrained_language_model_path='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m_tp2.nemo' \
3468+
model.language_model_path='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m_tp2.nemo' \
34633469
model.existing_tasks=[] \
34643470
model.new_tasks=['squad'] \
34653471
model.data.train_ds=['/home/TestData/nlp/prompt_learning/squad_CI_test.jsonl'] \
@@ -3469,13 +3475,15 @@ assert_frame_equal(training_curve, gt_curve, rtol=1e-3, atol=1e-3)"'''
34693475
sh "rm -rf /home/TestData/nlp/prompt_learning/t5_p_tuning_test_tp2"
34703476
sh "python examples/nlp/language_modeling/megatron_t5_prompt_learning_eval.py \
34713477
virtual_prompt_model_file='/home/TestData/nlp/prompt_learning/t5_p_tuning_test_tp2.nemo' \
3472-
pretrained_language_model_file='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m_tp2.nemo' \
3478+
language_model_path='/home/TestData/nlp/megatron_t5/8m/megatron_t5_8m_tp2.nemo' \
34733479
data.test_ds=['/home/TestData/nlp/prompt_learning/squad_CI_test.jsonl'] \
3480+
pred_file_path='/home/TestData/nlp/prompt_learning/t5_p_tuning_test_tp2_preds.txt' \
34743481
tensor_model_parallel_size=2 \
34753482
trainer.devices=2 \
34763483
data.global_batch_size=8 \
34773484
data.micro_batch_size=8"
34783485
sh "rm -rf /home/TestData/nlp/prompt_learning/t5_p_tuning_test_tp2.nemo"
3486+
sh "rm -rf /home/TestData/nlp/prompt_learning/t5_p_tuning_test_tp2_preds.txt"
34793487
}
34803488
}
34813489
}

README.rst

+3-9
Original file line numberDiff line numberDiff line change
@@ -200,16 +200,10 @@ Megatron GPT training requires NVIDIA Apex to be installed.
200200

201201
.. code-block:: bash
202202
203-
git clone https://github.com/NVIDIA/apex
203+
git clone https://github.com/ericharper/apex.git
204204
cd apex
205-
git checkout 3c19f1061879394f28272a99a7ea26d58f72dace
206-
pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" --global-option="--fast_layer_norm" ./
207-
208-
.. note::
209-
210-
You may need to modify [setup.py](https://github.com/NVIDIA/apex/blob/3c19f1061879394f28272a99a7ea26d58f72dace/setup.py) if
211-
your version of CUDA does not match the version used to compile Pytorch binaries, comment lines 33-41 in the above link
212-
before installing.
205+
git checkout nm_v1.11.0
206+
pip install -v --disable-pip-version-check --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" --global-option="--fast_layer_norm" --global-option="--distributed_adam" --global-option="--deprecated_fused_adam" ./
213207
214208
Docker containers:
215209
~~~~~~~~~~~~~~~~~~

docs/source/asr/data/benchmark_hr.csv

+3
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,3 @@
1+
Model,Model Base Class,Model Card
2+
stt_hr_conformer_ctc_large,EncDecCTCModel,"https://ngc.nvidia.com/catalog/models/nvidia:nemo:stt_hr_conformer_ctc_large"
3+
stt_hr_conformer_transducer_large,EncDecRNNTBPEModel,"https://ngc.nvidia.com/catalog/models/nvidia:nemo:stt_hr_conformer_transducer_large"
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,3 @@
1+
Model Name,Language,ParlaSpeech-HR v1.0 (dev),ParlaSpeech-HR v1.0 (test)
2+
stt_hr_conformer_ctc_large,hr,4.43,4.70
3+
stt_hr_conformer_transducer_large,hr,4.56,4.69

docs/source/asr/datasets.rst

+1-1
Original file line numberDiff line numberDiff line change
@@ -171,7 +171,7 @@ The audio files can be of any format supported by `Pydub <https://github.com/jia
171171
WAV files as they are the default and have been most thoroughly tested.
172172

173173
There should be one manifest file per dataset that will be passed in, therefore, if the user wants separate training and validation
174-
datasets, they should also have separate manifests. Otherwise, thay will be loading validation data with their training data and vice
174+
datasets, they should also have separate manifests. Otherwise, they will be loading validation data with their training data and vice
175175
versa.
176176

177177
Each line of the manifest should be in the following format:

docs/source/asr/models.rst

+12-6
Original file line numberDiff line numberDiff line change
@@ -132,15 +132,17 @@ Cache-aware Streaming Conformer
132132

133133
Buffered streaming uses overlapping chunks to make an offline ASR model to be used for streaming with reasonable accuracy. However, it uses significant amount of duplication in computations due to the overlapping chunks.
134134
Also there is a accuracy gep between the offline model and the streaming one as there is inconsistency between how we train the model and how we perform inference for streaming.
135-
The Cache-aware Streaming Conformer models would tackle and address these disadvantages. They are variants of Conformer which are trained with limited right context and it would make it possible to match the training and inference.
135+
The Cache-aware Streaming Conformer models would tackle and address these disadvantages. These streaming Conformers are trained with limited right context that it would make it possible to match how the model is being used in both the training and inference.
136+
They also uses caching to store intermediate activations to avoid any duplication in compute.
136137
The cache-aware approach is supported for both the Conformer-CTC and Conformer-Transducer and enables the model to be used very efficiently for streaming.
137138

138-
Three categories of layers in Conformer have access to right tokens: 1-depthwise convolutions 2-self-attention, and 3-convolutions in downsampling layers.
139+
Three categories of layers in Conformer have access to right tokens: 1-depthwise convolutions 2-self-attention, and 3-convolutions in the downsampling layers.
139140
Streaming Conformer models uses causal convolutions or convolutions with lower right context and also self-attention with limited right context to limit the effective right context for the input.
140-
The model trained with such limitations can be used in streaming mode and give the exact same output and accuracy as when the whole audio is given to the model in offline mode.
141+
The model trained with such limitations can be used in streaming mode and give the exact same outputs and accuracy as when the whole audio is given to the model in offline mode.
141142
These model can use caching mechanism to store and reuse the activations during streaming inference to avoid any duplications in the computations as much as possible.
142143

143144
We support the following three right context modeling:
145+
144146
* fully causal model with zero look-ahead: tokens would not see any future tokens. convolution layers are all causal and right tokens are masked for self-attention.
145147
It gives zero latency but with limited accuracy.
146148
To train such a model, you need to set `encoder.att_context_size=[left_context, 0]` and `encoder.conv_context_size=causal` in the config.
@@ -155,9 +157,9 @@ This approach is more efficient than regular look-ahead in terms of computations
155157
In terms of accuracy, this approach gives similar or even better results in term of accuracy than regular look-ahead as each token in each layer have access to more tokens on average. That is why we recommend to use this approach for streaming.
156158

157159

158-
** Note: Latencies are based on the assumption that the forward time of the network is zero.
160+
** Note: Latencies are based on the assumption that the forward time of the network is zero and it just estimates the time needed after a frame would be available until it is passed through the model.
159161

160-
Approaches with non-zero look-ahead can give significantly better accuracy by sacrificing latency. The latency can get controlled by the left context size.
162+
Approaches with non-zero look-ahead can give significantly better accuracy by sacrificing latency. The latency can get controlled by the left context size. Increasing the right context would help the accuracy to a limit but would increase the compuation time.
161163

162164

163165
In all modes, left context can be controlled by the number of tokens to be visible in the self-attention and the kernel size of the convolutions.
@@ -168,12 +170,16 @@ Left context of convolutions is dependent to the their kernel size while it can
168170
Self-attention left context of around 6 secs would give close result to have unlimited left context. For a model with 4x downsampling and shift window of 10ms in the preprocessor, each token corresponds to 4*10=40ms.
169171

170172
If striding approach is used for downsampling, all the convolutions in downsampling would be fully causal and don't see future tokens.
171-
It is recommended to use stacking for streaming model which is significantly faster and uses less memory.
173+
You may use stacking for downsampling in the streaming models which is significantly faster and uses less memory.
174+
It also does not some of the the limitations with striding and vggnet and you may use any downsampling rate.
172175

173176
You may find the example config files of cache-aware streaming Conformer models at
174177
``<NeMo_git_root>/examples/asr/conf/conformer/streaming/conformer_transducer_bpe_streaming.yaml`` for Transducer variant and
175178
at ``<NeMo_git_root>/examples/asr/conf/conformer/streaming/conformer_ctc_bpe.yaml`` for CTC variant.
176179

180+
To simulate cache-aware stremaing, you may use the script at ``<NeMo_git_root>/examples/asr/asr_streaming/speech_to_text_streaming_infer.py``. It can simulate streaming in single stream or multi-stream mode (in batches) for an ASR model.
181+
This script can be used for models trained offline with full-context but the accuracy would not be great unless the chunk size is large enough which would result in high latency.
182+
It is recommended to train a model in streaming model with limited context for this script. More info can be found in the script.
177183

178184
.. _LSTM-Transducer_model:
179185

docs/source/asr/scores.rst

+10
Original file line numberDiff line numberDiff line change
@@ -169,6 +169,16 @@ FR
169169

170170
--------------------
171171

172+
HR
173+
^^
174+
175+
.. csv-table::
176+
:header-rows: 1
177+
:align: left
178+
:file: data/scores/hr/conformer_hr.csv
179+
180+
--------------------
181+
172182
IT
173183
^^
174184

docs/source/conf.py

+1
Original file line numberDiff line numberDiff line change
@@ -120,6 +120,7 @@
120120
'nlp/text_normalization/tn_itn_all.bib',
121121
'tools/tools_all.bib',
122122
'tts_all.bib',
123+
'text_processing/text_processing_all.bib',
123124
'core/adapters/adapter_bib.bib',
124125
]
125126

docs/source/index.rst

+9
Original file line numberDiff line numberDiff line change
@@ -44,6 +44,7 @@ NVIDIA NeMo User Guide
4444
nlp/machine_translation/machine_translation
4545
nlp/text_normalization/intro
4646
nlp/api
47+
nlp/models
4748

4849

4950
.. toctree::
@@ -60,6 +61,14 @@ NVIDIA NeMo User Guide
6061
:caption: Common
6162
:name: Common
6263

64+
text_processing/intro
65+
66+
.. toctree::
67+
:maxdepth: 2
68+
:caption: Text Processing
69+
:name: Text Processing
70+
71+
text_processing/g2p/g2p
6372
common/intro
6473

6574

0 commit comments

Comments
 (0)