-
Notifications
You must be signed in to change notification settings - Fork 2.4k
4622 lines (4429 loc) · 211 KB
/
cicd-main.yml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# Copyright (c) 2020-2021, NVIDIA CORPORATION.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
name: "CICD NeMo"
on:
pull_request:
branches:
- 'main'
- 'r**'
types: [ labeled ]
concurrency:
group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
cancel-in-progress: true
jobs:
gpu-test:
runs-on: self-hosted-azure
if: ${{ github.event.label.name == 'Run CICD' }}
steps:
- name: Run nvidia-smi test
run: |
whoami
nvidia-smi
cicd-cluster-clean:
runs-on: self-hosted-azure-builder
if: ${{ github.event.label.name == 'Run CICD' }}
steps:
- name: Clean server from old files
run: |
docker container prune --filter "until=24h" --force
docker image prune -a --filter "until=24h" --force
cicd-test-container-setup:
needs: [cicd-cluster-clean]
runs-on: self-hosted-azure-builder
if: ${{ github.event.label.name == 'Run CICD' }}
steps:
- name: Checkout repository
uses: actions/checkout@v4
with:
path: ${{ github.run_id }}
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@v3
with:
# We use `docker` driver as this speeds things up for
# trivial (non-multi-stage) builds.
driver: docker
- name: Build and push
uses: docker/build-push-action@v5
with:
file: Dockerfile.ci
push: true
cache-from: nemoci.azurecr.io/nemo_container:latest
cache-to: type=inline
tags: |
nemoci.azurecr.io/nemo_container_${{ github.run_id }}
nemoci.azurecr.io/nemo_container:latest
- name: Run some checks
run: |
docker run --rm --device=/dev/nvidia0 --gpus all --shm-size=8g --env TRANSFORMERS_OFFLINE=0 --env HYDRA_FULL_ERROR=1 --env PYTHONUNBUFFERED=1 nemoci.azurecr.io/nemo_container_${{ github.run_id }} bash -c '\
# PyTorch Lightning version
python -c "import pytorch_lightning; print(pytorch_lightning.__version__)"
# PyTorch Lightning DDP Checks
CUDA_VISIBLE_DEVICES="0,1" python "tests/core_ptl/check_for_ranks.py"
# Basic Import Checks
python -c "import nemo.collections.asr as nemo_asr"
python -c "import nemo.collections.nlp as nemo_nlp"
python -c "import nemo.collections.tts as nemo_tts"
python setup.py style
python tests/check_copyright_header.py --dir .
# These checks are not crucial
exit 0
'
### \'\'
L0_Unit_Tests_GPU:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
TIMEOUT: 60
SCRIPT: |
NEMO_NUMBA_MINVER=0.53 pytest -m "not pleasefixme" --with_downloads
IS_OPTIONAL: true
L0_Unit_Tests_CPU:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-cpu
TIMEOUT: 60
SCRIPT: |
CUDA_VISIBLE_DEVICES="" NEMO_NUMBA_MINVER=0.53 pytest -m "not pleasefixme" --cpu --with_downloads --relax_numba_compat
L0_Setup_Test_Data_And_Models:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python -m tests.setup --save_dir /home/TestData/nlp
## - name: L2: Multimodal Imagen Train
# L2: Community LLM Checkpoints tests
L2_Community_LLM_Checkpoints_tests_Llama:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
CUDA_VISIBLE_DEVICES=0 python scripts/checkpoint_converters/convert_llama_hf_to_nemo.py \
--input_name_or_path=/home/TestData/nlp/megatron_llama/llama-ci-hf-tiny \
--output_path=/home/TestData/nlp/megatron_llama/llama_ci.nemo \
--precision=16
AFTER_SCRIPT: |
rm -rf /home/TestData/nlp/megatron_llama/model_weights
L2_Community_LLM_Checkpoints_tests_Llama3:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
CUDA_VISIBLE_DEVICES=0 python scripts/checkpoint_converters/convert_llama_hf_to_nemo.py \
--input_name_or_path=/home/TestData/nlp/megatron_llama/llama3-ci-hf \
--output_path=/home/TestData/nlp/megatron_llama/llama3-ci-hf/llama3_ci.nemo \
--precision=16
AFTER_SCRIPT: |
rm -f /home/TestData/nlp/megatron_llama/llama3-ci-hf/llama3_ci.nemo
rm -rf /home/TestData/nlp/megatron_llama/llama3-ci-hf/model_weights
L2_Community_LLM_Checkpoints_tests_StarCoder:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
mkdir -p /home/TestData/nlp/megatron_gpt/starcoder-ci-hf/${{ github.run_id }};
python scripts/checkpoint_converters/convert_starcoder_hf_to_nemo.py \
--input_name_or_path /home/TestData/nlp/megatron_gpt/starcoder-ci-hf \
--output_path /home/TestData/nlp/megatron_gpt/starcoder-ci-hf/${{ github.run_id }}
AFTER_SCRIPT: |
rm -rf /home/TestData/nlp/megatron_gpt/starcoder-ci-hf/megatron_starcoder_tp1_pp1.nemo;
rm -rf /home/TestData/nlp/megatron_gpt/starcoder-ci-hf/${{ github.run_id }}/
rm -rf /home/TestData/nlp/megatron_gpt/starcoder-ci-hf/model_weights
L2_Community_LLM_Checkpoints_tests_Falcon:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python scripts/checkpoint_converters/convert_falcon_hf_to_nemo.py \
--input_name_or_path /home/TestData/nlp/megatron_gpt/falcon-ci-hf \
--output_path /home/TestData/nlp/megatron_gpt/falcon-ci-hf/falcon_ci.nemo
rm -f /home/TestData/nlp/megatron_gpt/falcon-ci-hf/falcon_ci.nemo
AFTER_SCRIPT: |
rm -rf /home/TestData/nlp/megatron_gpt/falcon-ci-hf/model_weights
# this test is using a 7B model which is too large for GitHub CI
# replace the model in this test with a toy model or move the test
# to the nightly CI
# OPTIONAL_L2_Community_LLM_Checkpoints_tests_Baichuan2:
# needs: [cicd-test-container-setup]
# runs-on: self-hosted-azure
# container:
# image: nemoci.azurecr.io/nemo_container_${{ github.run_id }}
# options:
# # --user 0:128
# --device=/dev/nvidia0
# --gpus all
# --shm-size=8g
# --env TRANSFORMERS_OFFLINE=0
# --env HYDRA_FULL_ERROR=1
# --volume /mnt/datadrive/TestData:/home/TestData
# steps:
# - name: Checkout repository
# uses: actions/checkout@v4
# - run: |
# python scripts/checkpoint_converters/convert_baichuan2_hf_to_nemo.py \
# --input_name_or_path=/home/TestData/nlp/megatron_gpt/Baichuan2-7B-Base \
# --output_path=/home/TestData/nlp/megatron_gpt/Baichuan2-7B-Base/ci.nemo
# rm -f /home/TestData/nlp/megatron_gpt/Baichuan2-7B-Base/ci.nemo
# - uses: "NVIDIA/NeMo/.github/actions/cancel-workflow@main"
# if: "failure()"
L2_PTQ_Llama2_Export_Only:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/nlp/language_modeling/megatron_gpt_ptq.py \
model.restore_from_path=/home/TestData/nlp/megatron_llama/llama_ci.nemo \
quantization.algorithm=null \
export.save_path=/home/TestData/nlp/megatron_llama/ci_baseline
AFTER_SCRIPT: |
rm -rf /home/TestData/nlp/megatron_llama/ci_baseline
L2_PTQ_Llama2_FP8:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/nlp/language_modeling/megatron_gpt_ptq.py \
model.restore_from_path=/home/TestData/nlp/megatron_llama/llama_ci.nemo \
model.tensor_model_parallel_size=2 \
trainer.devices=2 \
quantization.calib_dataset=/home/TestData/nlp/test_quantization/test.json \
quantization.algorithm=fp8 \
quantization.num_calib_size=8 \
inference.batch_size=2 \
export.inference_tensor_parallel=2 \
export.save_path=/home/TestData/nlp/megatron_llama/ci_fp8.qnemo
AFTER_SCRIPT: |
rm -rf /home/TestData/nlp/megatron_llama/ci_fp8.qnemo
L2_PTQ_Llama2_INT8_SQ:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/nlp/language_modeling/megatron_gpt_ptq.py \
model.restore_from_path=/home/TestData/nlp/megatron_llama/llama_ci.nemo \
quantization.calib_dataset=/home/TestData/nlp/test_quantization/test.json \
quantization.algorithm=int8_sq \
quantization.num_calib_size=8 \
inference.batch_size=2 \
export.save_path=/home/TestData/nlp/megatron_llama/ci_int8_sq.qnemo
AFTER_SCRIPT: |
rm -rf /home/TestData/nlp/megatron_llama/ci_int8_sq.qnemo
# TODO: investigate int4_awq stuck issues and restore the test
#L2_PTQ_Llama2_INT4_AWQ:
# needs: [cicd-test-container-setup]
# runs-on: self-hosted-azure
# timeout-minutes: 10
# container:
# image: nemoci.azurecr.io/nemo_container_${{ github.run_id }}
# options:
# # --user 0:128
# --device=/dev/nvidia0
# --gpus all
# --shm-size=8g
# --env TRANSFORMERS_OFFLINE=0
# --env HYDRA_FULL_ERROR=1
# --volume /mnt/datadrive/TestData:/home/TestData
# steps:
# - name: Checkout repository
# uses: actions/checkout@v4
# - run: |
# python examples/nlp/language_modeling/megatron_gpt_ptq.py \
# model.restore_from_path=/home/TestData/nlp/megatron_llama/llama_ci.nemo \
# model.tensor_model_parallel_size=1 \
# trainer.devices=1 \
# quantization.calib_dataset=/home/TestData/nlp/test_quantization/test.json \
# quantization.algorithm=int4_awq \
# quantization.num_calib_size=8 \
# inference.batch_size=2 \
# export.save_path=/home/TestData/nlp/megatron_llama/ci_int4_awq.qnemo
#
# rm -rf /home/TestData/nlp/megatron_llama/ci_int4_awq.qnemo
#- uses: "NVIDIA/NeMo/.github/actions/cancel-workflow@main"
# if: "failure()"
L2_QAT_Llama2_INT4:
needs: [cicd-test-container-setup]
runs-on: self-hosted-azure
timeout-minutes: 10
container:
image: nemoci.azurecr.io/nemo_container_${{ github.run_id }}
options:
# --user 0:128
--device=/dev/nvidia0
--gpus all
--shm-size=8g
--env TRANSFORMERS_OFFLINE=0
--env HYDRA_FULL_ERROR=1
--volume /mnt/datadrive/TestData:/home/TestData
steps:
- name: Checkout repository
uses: actions/checkout@v4
- run: |
python examples/nlp/language_modeling/tuning/megatron_gpt_qat.py \
quantization.algorithm=int4 \
quantization.num_calib_size=8 \
trainer.devices=1 \
trainer.num_nodes=1 \
trainer.max_steps=4 \
trainer.val_check_interval=4 \
+trainer.limit_val_batches=2 \
exp_manager.explicit_log_dir=llama2_qat_results \
model.restore_from_path=/home/TestData/nlp/megatron_llama/llama_ci.nemo \
model.tensor_model_parallel_size=1 \
model.pipeline_model_parallel_size=1 \
model.global_batch_size=2 \
model.data.train_ds.file_names=[/home/TestData/nlp/megatron_sft/quarel.jsonl] \
model.data.train_ds.concat_sampling_probabilities=[1.0] \
model.data.validation_ds.file_names=[/home/TestData/nlp/megatron_sft/quarel.jsonl]
rm -rf llama2_qat_results
- uses: "NVIDIA/NeMo/.github/actions/cancel-workflow@main"
if: "failure()"
# L2: ASR dev run
ASR_dev_run_Speech_to_Text:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/asr_ctc/speech_to_text_ctc.py \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_to_text_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_results
ASR_dev_run_Speech_to_Text_WPE_-_CitriNet:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/asr_ctc/speech_to_text_ctc_bpe.py \
--config-path="../conf/citrinet/" --config-name="config_bpe" \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
model.tokenizer.dir="/home/TestData/asr_tokenizers/an4_wpe_128/" \
model.tokenizer.type="wpe" \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_to_text_wpe_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_wpe_results
ASR_dev_run_Speech_Pre-training_-_CitriNet:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/speech_pretraining/speech_pre_training.py \
--config-path="../conf/ssl/citrinet/" --config-name="citrinet_ssl_ci" \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_pre_training_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_pre_training_results
ASR_dev_run_Speech_To_Text_Finetuning:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/speech_to_text_finetune.py \
--config-path="conf/asr_finetune" --config-name="speech_to_text_finetune" \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
init_from_nemo_model=/home/TestData/asr/stt_en_fastconformer_transducer_large.nemo \
model.tokenizer.update_tokenizer=False \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_finetuning_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_finetuning_results
OPTIONAL_ASR_dev_run_Speech_To_Text_HF_Finetuning:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |-
python examples/asr/speech_to_text_finetune.py \
--config-path="conf/asr_finetune" --config-name="speech_to_text_hf_finetune" \
~model.train_ds.hf_data_cfg \
model.train_ds.num_workers=1 \
model.train_ds.batch_size=2 model.validation_ds.batch_size=2 \
model.train_ds.streaming=true \
+model.train_ds.hf_data_cfg.path="librispeech_asr" \
+model.train_ds.hf_data_cfg.name=null \
+model.train_ds.hf_data_cfg.split="test.clean" \
+model.train_ds.hf_data_cfg.streaming=true \
~model.validation_ds.hf_data_cfg \
model.validation_ds.streaming=true \
+model.validation_ds.hf_data_cfg.path="librispeech_asr" \
+model.validation_ds.hf_data_cfg.name=null \
+model.validation_ds.hf_data_cfg.split="test.clean" \
+model.validation_ds.hf_data_cfg.streaming=true \
~model.test_ds \
init_from_nemo_model=/home/TestData/asr/stt_en_fastconformer_transducer_large.nemo \
model.tokenizer.update_tokenizer=False \
model.optim.sched.warmup_steps=0 \
+model.optim.sched.max_steps=3 \
trainer.max_epochs=null \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_finetuning_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_finetuning_results
IS_OPTIONAL: true
ASR_dev_run_Speech_to_Text_WPE_-_Conformer:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/asr_ctc/speech_to_text_ctc_bpe.py \
--config-path="../conf/conformer" --config-name="conformer_ctc_bpe" \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
model.tokenizer.dir="/home/TestData/asr_tokenizers/an4_wpe_128/" \
model.tokenizer.type="wpe" \
model.train_ds.batch_size=4 \
model.validation_ds.batch_size=4 \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_to_text_wpe_conformer_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_wpe_conformer_results
# L2: ASR dev run - part two
ASR_dev_run-part_two_Speech_to_Text_WPE_-_Squeezeformer:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/asr_ctc/speech_to_text_ctc_bpe.py \
--config-path="../conf/squeezeformer" --config-name="squeezeformer_ctc_bpe" \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
model.tokenizer.dir="/home/TestData/asr_tokenizers/an4_wpe_128/" \
model.tokenizer.type="wpe" \
model.encoder.d_model=144 \
model.train_ds.batch_size=4 \
model.validation_ds.batch_size=4 \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_to_text_wpe_squeezeformer_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_wpe_squeezeformer_results
L2_Speech_to_Text_EMA:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/asr/asr_ctc/speech_to_text_ctc.py \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
trainer.devices=2 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
+exp_manager.ema.enable=True \
exp_manager.exp_dir=examples/asr/speech_to_text_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_results
# L2_Speech_to_Text_AED:
# needs: [cicd-test-container-setup]
# runs-on: self-hosted-azure-gpus-1
# container:
# image: nemoci.azurecr.io/nemo_container_${{ github.run_id }}
# options:
# # --user 0:128
# --device=/dev/nvidia0
# --gpus all
# --shm-size=8g
# --env TRANSFORMERS_OFFLINE=0
# --env HYDRA_FULL_ERROR=1
# --volume /mnt/datadrive/TestData:/home/TestData
# steps:
# - name: Checkout repository
# uses: actions/checkout@v4
# - run: |
# python examples/asr/speech_multitask/speech_to_text_aed.py \
# model.prompt_format=canary \
# model.model_defaults.asr_enc_hidden=256 \
# model.model_defaults.lm_dec_hidden=256 \
# model.encoder.n_layers=12 \
# model.transf_encoder.num_layers=0 \
# model.transf_decoder.config_dict.num_layers=12 \
# model.train_ds.manifest_filepath=/home/TestData/asr/manifests/canary/an4_canary_train.json \
# ++model.train_ds.is_tarred=false \
# model.train_ds.batch_duration=60 \
# +model.train_ds.text_field="answer" \
# +model.train_ds.lang_field="target_lang" \
# model.validation_ds.manifest_filepath=/home/TestData/asr/manifests/canary/an4_canary_val.json \
# +model.validation_ds.text_field="answer" \
# +model.validation_ds.lang_field="target_lang" \
# model.test_ds.manifest_filepath=/home/TestData/asr/manifests/canary/an4_canary_val.json \
# +model.test_ds.text_field="answer" \
# +model.test_ds.lang_field="target_lang" \
# model.tokenizer.langs.spl_tokens.dir=/home/TestData/asr_tokenizers/canary/canary_spl_tokenizer_v32 \
# model.tokenizer.langs.spl_tokens.type="bpe" \
# model.tokenizer.langs.en.dir=/home/TestData/asr_tokenizers/canary/en/tokenizer_spe_bpe_v1024_max_4 \
# model.tokenizer.langs.en.type=bpe \
# ++model.tokenizer.langs.es.dir=/home/TestData/asr_tokenizers/canary/es/tokenizer_spe_bpe_v1024_max_4 \
# ++model.tokenizer.langs.es.type=bpe \
# trainer.devices=1 \
# trainer.accelerator="gpu" \
# +trainer.use_distributed_sampler=false \
# +trainer.fast_dev_run=True \
# exp_manager.exp_dir=examples/asr/speech_to_text_aed_results
# rm -rf examples/asr/speech_to_text_results
# L2: Speaker dev run
L2_Speaker_dev_run_Speaker_Recognition:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/speaker_tasks/recognition/speaker_reco.py \
model.train_ds.batch_size=10 \
model.validation_ds.batch_size=2 \
model.train_ds.manifest_filepath=/home/TestData/an4_speaker/train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_speaker/dev.json \
model.decoder.num_classes=2 \
trainer.max_epochs=10 \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/speaker_tasks/recognition/speaker_recognition_results
AFTER_SCRIPT: |
rm -rf examples/speaker_tasks/recognition/speaker_recognition_results
L2_Speaker_dev_run_Speaker_Diarization:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/speaker_tasks/diarization/neural_diarizer/multiscale_diar_decoder.py \
model.diarizer.speaker_embeddings.model_path=titanet_large \
model.train_ds.batch_size=5 \
model.validation_ds.batch_size=5 \
model.train_ds.emb_dir=examples/speaker_tasks/diarization/speaker_diarization_results \
model.validation_ds.emb_dir=examples/speaker_tasks/diarization/speaker_diarization_results \
model.train_ds.manifest_filepath=/home/TestData/an4_diarizer/simulated_train/msdd_data.50step.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_diarizer/simulated_valid/msdd_data.50step.json \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/speaker_tasks/diarization/speaker_diarization_results
AFTER_SCRIPT: |
rm -rf examples/speaker_tasks/diarization/speaker_diarization_results
L2_Speaker_dev_run_Speech_to_Label:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/speech_classification/speech_to_label.py \
model.train_ds.manifest_filepath=/home/TestData/speech_commands/train_manifest.json \
model.validation_ds.manifest_filepath=/home/TestData/speech_commands/test_manifest.json \
model.test_ds.manifest_filepath=/home/TestData/speech_commands/test_manifest.json \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
model.preprocessor._target_=nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor \
~model.preprocessor.window_size \
~model.preprocessor.window_stride \
~model.preprocessor.window \
~model.preprocessor.n_mels \
~model.preprocessor.n_mfcc \
~model.preprocessor.n_fft \
exp_manager.exp_dir=examples/asr/speech_to_label_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_label_results
L2_Speaker_dev_run_Speaker_Diarization_with_ASR_Inference:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/speaker_tasks/diarization/clustering_diarizer/offline_diar_with_asr_infer.py \
diarizer.manifest_filepath=/home/TestData/an4_diarizer/an4_manifest.json \
diarizer.speaker_embeddings.model_path=/home/TestData/an4_diarizer/spkr.nemo \
diarizer.speaker_embeddings.parameters.save_embeddings=True \
diarizer.speaker_embeddings.parameters.window_length_in_sec=[1.5] \
diarizer.speaker_embeddings.parameters.shift_length_in_sec=[0.75] \
diarizer.speaker_embeddings.parameters.multiscale_weights=[1.0] \
diarizer.asr.model_path=QuartzNet15x5Base-En \
diarizer.asr.parameters.asr_based_vad=True \
diarizer.out_dir=examples/speaker_tasks/diarization/speaker_diarization_asr_results
AFTER_SCRIPT: |
rm -rf examples/speaker_tasks/diarization/speaker_diarization_asr_results
L2_Speaker_dev_run_Clustering_Diarizer_Inference:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/speaker_tasks/diarization/clustering_diarizer/offline_diar_infer.py \
diarizer.manifest_filepath=/home/TestData/an4_diarizer/an4_manifest.json \
diarizer.speaker_embeddings.model_path=/home/TestData/an4_diarizer/spkr.nemo \
diarizer.speaker_embeddings.parameters.save_embeddings=True \
diarizer.speaker_embeddings.parameters.window_length_in_sec=1.5 \
diarizer.speaker_embeddings.parameters.shift_length_in_sec=0.75 \
diarizer.speaker_embeddings.parameters.multiscale_weights=null \
diarizer.vad.model_path=/home/TestData/an4_diarizer/MatchboxNet_VAD_3x2.nemo \
diarizer.out_dir=examples/speaker_tasks/diarization/clustering_diarizer_results
AFTER_SCRIPT: |
rm -rf examples/speaker_tasks/diarization/clustering_diarizer_results
L2_Speaker_dev_run_Neural_Diarizer_Inference:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/speaker_tasks/diarization/neural_diarizer/multiscale_diar_decoder_infer.py \
diarizer.manifest_filepath=/home/TestData/an4_diarizer/an4_manifest.json \
diarizer.msdd_model.model_path=/home/TestData/an4_diarizer/diar_msdd_telephonic.nemo \
diarizer.speaker_embeddings.parameters.save_embeddings=True \
diarizer.vad.model_path=/home/TestData/an4_diarizer/MatchboxNet_VAD_3x2.nemo \
diarizer.out_dir=examples/speaker_tasks/diarization/neural_diarizer_results
AFTER_SCRIPT: |
rm -rf examples/speaker_tasks/diarization/neural_diarizer_results
L2_Speaker_dev_run_Multispeaker_ASR_Data_Simulation:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python tools/speech_data_simulator/multispeaker_simulator.py \
--config-path=conf --config-name=data_simulator.yaml \
data_simulator.random_seed=42 \
data_simulator.manifest_filepath=/home/TestData/LibriSpeechShort/dev-clean-align-short.json \
data_simulator.outputs.output_dir=./test_simulator \
data_simulator.session_config.num_sessions=2 \
data_simulator.session_config.session_length=60
AFTER_SCRIPT: |
rm -rf ./test_simulator
# L2: ASR Multi-dataloader dev run
L2_ASR_Multi-dataloader_dev_run_Speech_to_Text_multi-dataloader:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/asr_ctc/speech_to_text_ctc.py \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=[/home/TestData/an4_dataset/an4_val.json,/home/TestData/an4_dataset/an4_val.json] \
trainer.devices=1 \
trainer.accelerator="gpu" \
trainer.max_epochs=1 \
trainer.max_steps=1 \
+trainer.num_sanity_val_steps=1 \
exp_manager.exp_dir=examples/asr/speech_to_text_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_results
L2_ASR_Multi-dataloader_dev_run_Speech_to_Label_multi-dataloader:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/speech_classification/speech_to_label.py \
model.train_ds.manifest_filepath=/home/TestData/speech_commands/train_manifest.json \
model.validation_ds.manifest_filepath=[/home/TestData/speech_commands/test_manifest.json,/home/TestData/speech_commands/test_manifest.json] \
trainer.devices=1 \
trainer.accelerator="gpu" \
trainer.max_epochs=1 \
trainer.max_steps=1 \
+trainer.num_sanity_val_steps=1 \
model.preprocessor._target_=nemo.collections.asr.modules.AudioToMelSpectrogramPreprocessor \
~model.preprocessor.window_size \
~model.preprocessor.window_stride \
~model.preprocessor.window \
~model.preprocessor.n_mels \
~model.preprocessor.n_mfcc \
~model.preprocessor.n_fft \
exp_manager.exp_dir=examples/asr/speech_to_label_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_label_results
# L2: ASR Adapters
L2_ASR_Adapters_Linear_Adapters:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/asr_adapters/train_asr_adapter.py \
model.pretrained_model="stt_en_conformer_ctc_small" \
model.adapter.adapter_name="an4" \
model.adapter.linear.in_features=176 \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
trainer.max_steps=5 \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_to_text_adapters_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_adapters_results
L2_ASR_Adapters_RelPos_MHA_Adapters:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
python examples/asr/asr_adapters/train_asr_adapter.py \
model.pretrained_model="stt_en_conformer_ctc_small" \
model.adapter.adapter_name="encoder:an4" \
model.adapter.adapter_type="tiny_attn" \
model.adapter.tiny_attn.n_feat=176 \
model.train_ds.manifest_filepath=/home/TestData/an4_dataset/an4_train.json \
model.validation_ds.manifest_filepath=/home/TestData/an4_dataset/an4_val.json \
trainer.max_steps=5 \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=True \
exp_manager.exp_dir=examples/asr/speech_to_text_adapters_mha_results
AFTER_SCRIPT: |
rm -rf examples/asr/speech_to_text_adapters_mha_results
# L2: Speech Transcription
L2_Speech_Transcription_Speech_to_Text_Transcribe:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
python examples/asr/transcribe_speech.py \
pretrained_name="QuartzNet15x5Base-En" \
audio_dir="/home/TestData/an4_transcribe/test_subset/" \
output_filename="stt_test_res.json" \
amp=true
AFTER_SCRIPT: |
rm -rf stt_test_res.json
# L2: Transducer alignment
L2_Transducer_alignment_Running_pytest:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
pytest tests/collections/asr/decoding/rnnt_alignments_check.py --durations=-1
# L2: Segmentation Tool
L2_Segmentation_Tool_Parallel_ctc_segmentation_test_L2_Eng_CitriNet_with_wav:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
cd tools/ctc_segmentation && \
TIME=`date +"%Y-%m-%d-%T"` && \
/bin/bash run_segmentation.sh \
--MODEL_NAME_OR_PATH="stt_en_citrinet_512_gamma_0_25" \
--DATA_DIR=/home/TestData/ctc_segmentation/eng \
--OUTPUT_DIR=/home/TestData/ctc_segmentation/eng/output${TIME} \
--LANGUAGE=en \
--USE_NEMO_NORMALIZATION="TRUE" && \
python /home/TestData/ctc_segmentation/verify_alignment.py \
-r /home/TestData/ctc_segmentation/eng/eng_valid_segments_1.7.txt \
-g /home/TestData/ctc_segmentation/eng/output${TIME}/verified_segments/nv_test_segments.txt;
AFTER_SCRIPT: |
rm -rf /home/TestData/ctc_segmentation/eng/output${TIME}
L2_Segmentation_Tool_Parallel_ctc_segmentation_test_L2_Ru_QN_with_mp3:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
cd tools/ctc_segmentation && \
TIME=`date +"%Y-%m-%d-%T"` && \
/bin/bash run_segmentation.sh \
--MODEL_NAME_OR_PATH=/home/TestData/ctc_segmentation/QuartzNet15x5-Ru-e512-wer14.45.nemo \
--DATA_DIR=/home/TestData/ctc_segmentation/ru \
--OUTPUT_DIR=/home/TestData/ctc_segmentation/ru/output${TIME} \
--LANGUAGE=ru \
--ADDITIONAL_SPLIT_SYMBOLS=";" && \
python /home/TestData/ctc_segmentation/verify_alignment.py \
-r /home/TestData/ctc_segmentation/ru/valid_ru_segments_1.7.txt \
-g /home/TestData/ctc_segmentation/ru/output${TIME}/verified_segments/ru_segments.txt;
rm -rf /home/TestData/ctc_segmentation/eng/output${TIME}
# L2: G2P Models
L2_G2P_Models_G2P_Conformer_training_evaluation_and_inference:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
cd examples/tts/g2p && \
TIME=`date +"%Y-%m-%d-%T"` && OUTPUT_DIR_CONFORMER=output_ctc_${TIME} && \
python g2p_train_and_evaluate.py \
train_manifest=/home/TestData/g2p/g2p.json \
validation_manifest=/home/TestData/g2p/g2p.json \
model.test_ds.manifest_filepath=/home/TestData/g2p/g2p.json \
model.tokenizer.dir=/home/TestData/g2p/tokenizer_spe_unigram_v512 \
trainer.max_epochs=1 \
model.max_source_len=64 \
trainer.devices=1 \
do_training=True \
do_testing=True \
exp_manager.exp_dir=${OUTPUT_DIR_CONFORMER} \
+exp_manager.use_datetime_version=False\
+exp_manager.version=test \
--config-name=g2p_conformer_ctc && \
python g2p_inference.py \
pretrained_model=${OUTPUT_DIR_CONFORMER}/G2P-Conformer-CTC/test/checkpoints/G2P-Conformer-CTC.nemo \
manifest_filepath=/home/TestData/g2p/g2p.json \
phoneme_field=text
# TODO: pleasefixme @redoctopus
# - name: ByT5G2P training, evaluation and inference
# run: |
# cd examples/tts/g2p && \
# TIME=`date +"%Y-%m-%d-%T"` && OUTPUT_DIR_T5=output_byt5_${TIME} && \
# python g2p_train_and_evaluate.py \
# train_manifest=/home/TestData/g2p/g2p.json \
# validation_manifest=/home/TestData/g2p/g2p.json \
# model.test_ds.manifest_filepath=/home/TestData/g2p/g2p.json \
# trainer.max_epochs=1 \
# model.max_source_len=64 \
# trainer.devices=1 \
# do_training=True \
# do_testing=True \
# exp_manager.exp_dir=${OUTPUT_DIR_T5} \
# +exp_manager.use_datetime_version=False\
# +exp_manager.version=test && \
# python g2p_inference.py \
# pretrained_model=${OUTPUT_DIR_T5}/T5G2P/test/checkpoints/T5G2P.nemo \
# manifest_filepath=/home/TestData/g2p/g2p.json \
# phoneme_field=text
# }
# }
# - uses: "NVIDIA/NeMo/.github/actions/cancel-workflow@main"
# if: "failure()"
L2_G2P_Models_HeteronymClassificationModel_training_evaluation_and_inference:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
cd examples/tts/g2p && \
TIME=`date +"%Y-%m-%d-%T"` && OUTPUT_DIR=output_${TIME} && \
python g2p_heteronym_classification_train_and_evaluate.py \
train_manifest=/home/TestData/g2p/manifest.json \
validation_manifest=/home/TestData/g2p/manifest.json \
test_manifest=/home/TestData/g2p/manifest.json \
model.wordids=/home/TestData/g2p/wordids.tsv \
trainer.max_epochs=1 \
model.max_seq_length=64 \
do_training=True \
do_testing=True \
exp_manager.exp_dir=${OUTPUT_DIR} \
+exp_manager.use_datetime_version=False\
+exp_manager.version=test && \
python g2p_heteronym_classification_inference.py \
manifest=/home/TestData/g2p/manifest.json \
pretrained_model=${OUTPUT_DIR}/HeteronymClassification/test/checkpoints/HeteronymClassification.nemo \
output_manifest=preds.json
# L2: Duplex Text Normalization
L2_Duplex_Text_Normalization_with_Tarred_dataset:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure
SCRIPT: |
cd examples/nlp/duplex_text_normalization && \
python duplex_text_normalization_train.py \
data.validation_ds.data_path=/home/TestData/nlp/duplex_text_norm/small_test.tsv \
mode=tn \
lang=en \
tagger_model.do_training=false \
decoder_model.transformer=t5-small \
data.validation_ds.batch_size=2 \
data.train_ds.use_cache=false \
data.validation_ds.use_cache=false \
data.test_ds.batch_size=2 \
data.train_ds.decoder_data_augmentation=false \
data.train_ds.num_workers=2 \
decoder_trainer.devices=[0,1] \
decoder_trainer.accelerator="gpu" \
data.train_ds.use_tarred_dataset=true \
+decoder_trainer.fast_dev_run=true \
decoder_exp_manager.create_checkpoint_callback=false \
data.train_ds.tar_metadata_file=/home/TestData/nlp/duplex_text_norm/tarred_small/metadata.json \
data.test_ds.use_cache=false \
data.test_ds.data_path=/home/TestData/nlp/duplex_text_norm/small_test.tsv
# L2: Intent and Slot Classification Tasks
L2_Intent_and_Slot_Classification_Tasks_Intent_and_Slot_Classification:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
cd examples/nlp/intent_slot_classification && \
python intent_slot_classification.py \
model.data_dir=/home/TestData/nlp/retail \
model.validation_ds.prefix=dev \
model.test_ds.prefix=dev \
trainer.devices=1 \
trainer.accelerator="gpu" \
+trainer.fast_dev_run=true \
exp_manager.exp_dir=checkpoints
AFTER_SCRIPT: |
rm -rf checkpoints
L2_Intent_and_Slot_Classification_Tasks_Multi-Label_Intent_and_Slot_Classification:
needs: [cicd-test-container-setup]
uses: ./.github/workflows/_test_template.yml
with:
RUNNER: self-hosted-azure-gpus-1
SCRIPT: |
cd examples/nlp/intent_slot_classification && \
python multi_label_intent_slot_classification.py \
model.data_dir=/home/TestData/nlp/new_multiatis \
model.validation_ds.prefix=dev \
model.test_ds.prefix=dev \
trainer.devices=1 \
+trainer.fast_dev_run=true \
exp_manager.exp_dir=checkpoints2
AFTER_SCRIPT: |
rm -rf checkpoints2
# TODO: add when megatron-bert is supported again
# stage('L2: Model Parallel Size 2 Megatron Text Classification') {
# when {
# anyOf{
# branch 'main'
# changeRequest target: 'main'
# }
# }
# failFast true
# steps{
# cd examples/nlp/text_classification && \
# python text_classification_with_bert.py \
# trainer.devices=[0,1] \
# trainer.accelerator="gpu" \
# trainer.num_nodes=1 \
# trainer.precision=16 \
# trainer.gradient_clip_val=1.0 \
# +trainer.fast_dev_run=true \
# model.dataset.num_classes=6 \
# model.train_ds.file_path=/home/TestData/nlp/retail_text_classification/train.tsv \
# model.train_ds.batch_size=4 \
# model.language_model.pretrained_model_name=megatron-bert-uncased \