Datasets:

---

renwei2024

/

nemo-github-issues

like 0

[]

[]

[ "Cannot root cause why DCO fails" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "This pull request **introduces 1 alert** when merging 4a0ca9b202c7c01f9d827cf392f0cd7cbb3af566 into 5d80627ca696384dc16e462f1e17efd33917a9dd - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-119caa9229f75b6c349ae1b681724d504f3e2940)\n\n**new alerts:**\n\n* 1 for Unused local variable" ]

[]

[]

[ "@xaiguy Can you please clarify what do you mean by \"custom NeMo model\"? \r\nEvery TLT model can be exported to Jarvis. But NeMo contains more models than TLT, including many \"research\" ones which can't be deployed to Jarvis yet.\r\n \r\nWe are constantly increasing the number of models supported by Jarvis by adding them to TLT.\r\n", "@okuchaiev Oh, pardon the ambiguity. With \"custom\" I just meant models fine-tuned with NeMo, in particular Jasper. I just mentioned TLT because I saw that it's very easy to save and export models to .ejvrs format from there and I was hoping to find something similar for NeMo.", "I saw a new passage was added to the docs, thanks! I'll close this issue now, as the answer can be found following the link below:\r\n\r\nhttps://docs.nvidia.com/deeplearning/jarvis/user-guide/docs/model-overview.html?highlight=servicemaker#nemo-to-jarvis" ]

[]

[]

[]

[]

[]

[ "Wrong branch" ]

[]

[]

[ "This pull request **introduces 10 alerts** and **fixes 6** when merging f44c492473364106bcc445a9b723ed3c961f2ac3 into cf23c451ecdc8bd8b0d2bd48103717121780e989 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-0671fc7ef505fcdfab6011715546b767428230d0)\n\n**new alerts:**\n\n* 9 for Unused import\n* 1 for Unnecessary pass\n\n**fixed alerts:**\n\n* 6 for Unused import" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "if the encoders & decoders are basically the same thing, would it make sense to have a single file that they import blocks from?", "> if the encoders & decoders are basically the same thing, would it make sense to have a single file that they import blocks from?\r\n\r\nThat is already happening to a good extent - both import MultiHeadAttention and PositionWiseFF blocks and that logic lies outside the Encoder/Decoder layers.", "This pull request **introduces 1 alert** when merging 21a598eb581a78f1dfa171dc1b41d2cb7ca21732 into a33d2794cba9cff5b29044f787e41f654325424a - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-afa1324fdce133834d87c84752630c214b1d11cb)\n\n**new alerts:**\n\n* 1 for Unused import", "LGTM", "EDIT: it's the same return issue i pointed out.\r\n\r\n\r\ndont merge yet; i am getting some sort of error:\r\n\r\n```\r\nTraceback (most recent call last):\r\n File \"examples/nlp/machine_translation/enc_dec_nmt.py\", line 127, in main\r\n trainer.fit(mt_model)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/trainer/trainer.py\", line 513, in fit\r\n self.dispatch()\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/trainer/trainer.py\", line 553, in dispatch\r\n self.accelerator.start_training(self)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/accelerators/accelerator.py\", line 74, in start_training\r\n self.training_type_plugin.start_training(trainer)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/plugins/training_type/training_type_plugin.py\", line 111, in start_training\r\n self._results = trainer.run_train()\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/trainer/trainer.py\", line 614, in run_train\r\n self.run_sanity_check(self.lightning_module)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/trainer/trainer.py\", line 863, in run_sanity_check\r\n _, eval_results = self.run_evaluation(max_batches=self.num_sanity_val_batches)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/trainer/trainer.py\", line 732, in run_evaluation\r\n output = self.evaluation_loop.evaluation_step(batch, batch_idx, dataloader_idx)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/trainer/evaluation_loop.py\", line 164, in evaluation_step\r\n output = self.trainer.accelerator.validation_step(args)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/accelerators/accelerator.py\", line 178, in validation_step\r\n return self.training_type_plugin.validation_step(*args)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/plugins/training_type/ddp.py\", line 290, in validation_step\r\n return self.model(*args, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 727, in _call_impl\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/parallel/distributed.py\", line 671, in forward\r\n output = self.module(*inputs[0], **kwargs[0])\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 727, in _call_impl\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/pytorch_lightning/overrides/base.py\", line 63, in forward\r\n output = self.module.validation_step(*inputs, **kwargs)\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/collections/nlp/models/machine_translation/mt_enc_dec_model.py\", line 234, in validation_step\r\n return self.eval_step(batch, batch_idx, 'val')\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/collections/nlp/models/machine_translation/mt_enc_dec_model.py\", line 199, in eval_step\r\n log_probs = self(src_ids, src_mask, tgt_ids, tgt_mask)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 727, in _call_impl\r\n result = self.forward(*input, **kwargs)\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/core/classes/common.py\", line 721, in __call__\r\n return wrapped(*args, **kwargs)\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/collections/nlp/models/machine_translation/mt_enc_dec_model.py\", line 167, in forward\r\n src_hiddens = self.encoder(src, src_mask)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 727, in _call_impl\r\n result = self.forward(*input, **kwargs)\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/collections/nlp/modules/common/transformer/transformer.py\", line 103, in forward\r\n encoder_hidden_states = self._encoder(embeddings, encoder_mask)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 727, in _call_impl\r\n result = self.forward(*input, **kwargs)\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/collections/nlp/modules/common/transformer/transformer_encoders.py\", line 158, in forward\r\n encoder_states = layer(encoder_states, encoder_attn_mask, memory_states)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 727, in _call_impl\r\n result = self.forward(*input, **kwargs)\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/collections/nlp/modules/common/transformer/transformer_encoders.py\", line 99, in forward\r\n self.forward_preln(encoder_query, encoder_mask, encoder_keys)\r\n File \"/home/dcg-adlr-mchrzanowski-source/NeMo2/nemo/collections/nlp/modules/common/transformer/transformer_encoders.py\", line 70, in forward_preln\r\n encoder_query = self.layer_norm_1(encoder_query)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/module.py\", line 727, in _call_impl\r\n result = self.forward(*input, **kwargs)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/modules/normalization.py\", line 170, in forward\r\n input, self.normalized_shape, self.weight, self.bias, self.eps)\r\n File \"/opt/conda/lib/python3.6/site-packages/torch/nn/functional.py\", line 2072, in layer_norm\r\n torch.backends.cudnn.enabled)\r\nTypeError: layer_norm(): argument 'input' (position 1) must be Tensor, not NoneType\r\n```\r\n\r\nsomewhere in between 3676721 and 9efb1e7", "Good catch @mchrzanowski . I would have thought one of our tests would have failed for something like this.", "Nice catch, our CI runs with`pre_ln=false` and this was happening only when `pre_ln=true`. Fixed now." ]

[ { "id": 1485815557, "node_id": "MDU6TGFiZWwxNDg1ODE1NTU3", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/bug", "name": "bug", "color": "d73a4a", "default": true, "description": "Something isn't working" } ]

[]

[ "I think we haven't properly implemented export for Conformer. @VahidooX to comment further. If this isn't supported yet, maybe should raise \"NotImplemented\" exception? ", "We have not tested the export for Conformer yet. I would raise NotImplemented for now. \r\nUpdate: I fixed the issue with the export function in this PR https://github.com/NVIDIA/NeMo/pull/1898. It is just a part of that PR, and not merged yet. Now it creates the onnx file but there is no guarantee that it works. yet", "closing since PR is merged" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ { "id": 1719393562, "node_id": "MDU6TGFiZWwxNzE5MzkzNTYy", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/feature", "name": "feature", "color": "68F3F0", "default": false, "description": "request/PR for a new feature" } ]

[ { "login": "okuchaiev", "id": 22035961, "node_id": "MDQ6VXNlcjIyMDM1OTYx", "avatar_url": "https://avatars.githubusercontent.com/u/22035961?v=4", "gravatar_id": "", "url": "https://api.github.com/users/okuchaiev", "html_url": "https://github.com/okuchaiev", "followers_url": "https://api.github.com/users/okuchaiev/followers", "following_url": "https://api.github.com/users/okuchaiev/following{/other_user}", "gists_url": "https://api.github.com/users/okuchaiev/gists{/gist_id}", "starred_url": "https://api.github.com/users/okuchaiev/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/okuchaiev/subscriptions", "organizations_url": "https://api.github.com/users/okuchaiev/orgs", "repos_url": "https://api.github.com/users/okuchaiev/repos", "events_url": "https://api.github.com/users/okuchaiev/events{/privacy}", "received_events_url": "https://api.github.com/users/okuchaiev/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "I also had to fix torch to 1.7.1. I also opened a pull request for that, but is not accepted to stick the version of torch. You can read more here https://github.com/NVIDIA/NeMo/issues/1868 and follow until you see my pull request, which is a temporary workaround for me. Maybe you should apply to yourself the same practice.", "Closing due to inactivity. Please reopen if still having the issue." ]

[]

[]

[]

[]

[]

[ "This pull request **introduces 9 alerts** when merging 8abd6c6c4439cc5ce3e1e65f5dfb1950faecceaf into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-57eaca60590a915a6de70d40a43478f84286be8b)\n\n**new alerts:**\n\n* 9 for Unused import", "This pull request **introduces 8 alerts** when merging 25c47c98b4f926d7ecde0983fd78afa298fa6262 into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-93a20e9ba45f5b25b0d1aea9927876859d0dc06b)\n\n**new alerts:**\n\n* 8 for Unused import", "This pull request **introduces 8 alerts** when merging 451c58ede732b03459401125db3d863ee3cde7d6 into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-7b59b100be33fa5ec49e9830cfd5a50e2564af38)\n\n**new alerts:**\n\n* 8 for Unused import", "This pull request **introduces 8 alerts** when merging bccf4359ded0623538005f8306e3e43aa0ee1901 into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-e64814b5067e63dd3def67107ec3ade6870ce1c3)\n\n**new alerts:**\n\n* 8 for Unused import", "This pull request **introduces 8 alerts** when merging e497e9c9f3cca1bc222eab775384894e52892100 into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-5ce35f5f9af6e8021615d26d34756d2a441f8444)\n\n**new alerts:**\n\n* 8 for Unused import", "This pull request **introduces 8 alerts** when merging e9e24b0642ba4442d8ecc1ed232cb6d0d22e5942 into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-707978e222e3dd83f9122e0d7f8a1868a5228ccf)\n\n**new alerts:**\n\n* 8 for Unused import", "This pull request **introduces 9 alerts** when merging ace6054f6ff8e187a3dabb6c8579ffe37994ed67 into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-4c85b3a77bb89c17a75099d929758ede363b0712)\n\n**new alerts:**\n\n* 9 for Unused import", "This pull request **introduces 9 alerts** when merging 840d85a2be80a1bdaaaa4ce89dac9e8d66ab210e into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-5448b514e9884d39d1bf7390ab6baa7cd4ee243f)\n\n**new alerts:**\n\n* 9 for Unused import", "This pull request **introduces 9 alerts** when merging b5548e8ac629d09238907accd078968e0eb2343c into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-bf6f33958e2a39abcf23e1556be4677f0ecfba77)\n\n**new alerts:**\n\n* 9 for Unused import", "This pull request **introduces 9 alerts** when merging 4ec5233841f60524fe467e0ef47efa67c38af3df into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-64b32681122ebce805dac4871d3c8ff5e86b9e20)\n\n**new alerts:**\n\n* 9 for Unused import", "This pull request **introduces 9 alerts** when merging 9415668530a3b8cecc6c59f8dfc32d614e8a710b into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-e8cf742f834fe689da5d110aae105f285ae04b5c)\n\n**new alerts:**\n\n* 9 for Unused import", "This pull request **introduces 9 alerts** when merging ce91bc6f760ec8d0cb5ceb6b20e77cb9b25d833b into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2d7e75c8d23771faac96c610a683be97be76f6a7)\n\n**new alerts:**\n\n* 9 for Unused import", "This pull request **introduces 10 alerts** when merging f716c2a62119dffda6970b180aaf8f2ac9da7297 into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-4dc03d91b173b523df4671949257c290a3719471)\n\n**new alerts:**\n\n* 10 for Unused import", "This pull request **introduces 1 alert** when merging c1ffc2f4afa9944f6e144c85cb1f7bdcc0746c90 into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-c99f08d69fe3218ec44d0ee9aa9b9c7bd82b2835)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 8c66456ce66de0e9ca0c1e59b333a35f531140e2 into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2c936e8bba47f81569faf00f09ed12652cb3500e)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 4fe9ec3856069c022b594cc05467f9dcde063bd6 into a33d2794cba9cff5b29044f787e41f654325424a - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2611a24d4bcd182d90cfe48ba77f8de8a98c5443)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging d0c0ad094f0d962bf46c3a0710f3d6fabdf89cd7 into a33d2794cba9cff5b29044f787e41f654325424a - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-faeb51f6d07a71183a9f2de39e302f23d2dcc555)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 2bf7978ace5bf9e1e9915e2aebd66247719e4da1 into 5d80627ca696384dc16e462f1e17efd33917a9dd - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-db259f4d4535a335027996896f845157461243e0)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging ac50ff7f38dc0b40fadd043ba89f48115432326c into 5d80627ca696384dc16e462f1e17efd33917a9dd - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-1ff35793b6715f4e03900e37765fde7ff64b2152)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 6d9d547a3f2a0c46e4d17d515d5bbc3da0872a87 into 05eb2587a95e512b5e95705d9f77582634abbd95 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-da27fb011b14e210486260eec93f4e64e784c32c)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 1a8aa0967db281d5e09a9efd400936b0e2eac727 into 646aff11aa788b7d874dae930ce6ffcd658c1f16 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-083514a3c31f38a8f876cbcabe0fcdef74f98916)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging e0cca8f60c7a3946eee8bd18010029bf33384d1a into 646aff11aa788b7d874dae930ce6ffcd658c1f16 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-814ada8b11f65d71e003845db1a54f0de081bb01)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 1876f3f6f1cd743ef30df8e34e935730892652a3 into b0f9f94469b1b7bacb10fb1851f8fabfd8b4323d - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-0e99d19428b41e342b29946e4215b117bcc5121e)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** and **fixes 1** when merging ff3b03ada6ec8677b8d4295fbea0d05138103fd5 into 4b8dd8ba7d02fd52e863f554b98f6be4c470948d - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-b5cebddab2c1ffcba131a6c0751a71c8dec470f9)\n\n**new alerts:**\n\n* 1 for Unused import\n\n**fixed alerts:**\n\n* 1 for Unused import" ]

[]

[]

[ "I did not played enough with that, but my first impression it was that with my studio microphone without dynamic compression and leveling, already was worst. Therefore at the beginning I will use with hardware dynamic compression. Once I am sure that this is the biggest issue, then I will make this compression in software also.", "ok thanks for replying. Now i need to use NeMo for audio recorded using phone /laptop microphone. Is there any way to improve the audio quality before giving it to the model. how to get the quality speech to text with phone recording /zoom audio ?", "Before going into development... first try to do some experiments. Take some audio processing tools, where you can do dynamic compression, limiting, even better if you can do proper noise reduction. Prepare manually a few variants of quality from the same file and compare the results. Then you will have the correct observation what makes sense to do and what doesn't make sense.\r\nIf you see the clear process what works, then the same thing can be implemented with librosa. (Of course some AI based noise reduction which you can make by iZotope RX you will not be able to reproduct by just using librosa.) Then you will go to search implementation examples from librosa or ISMIR, etc.\r\nThis requires a bit of work like this, a kind of R&D approach.", "@kruthikakr Could you share some samples that you are seeing an issue with? Most of our training data consists of clean studio recordings but you should get reasonable transcripts for noisy audio say up to 20 dB SNR. Could you also share which model/models you experimented with? \r\n\r\n> Now i need to use NeMo for audio recorded using phone /laptop microphone. Is there any way to improve the audio quality before giving it to the model. how to get the quality speech to text with phone recording /zoom audio ?\r\n\r\nYou could finetune the model to with noise augmentation to improve noise robustness of the model, we have a noise robust QuartzNet model that you could give a try. We have a new release coming up (in a couple of weeks) with better models trained on more conversational data that may fit your application.\r\n", "**1. we tried with 1 zoom recorded audio which was m4a format and the the size was7.62MB and then converted in to .wav format (15 MB size) which is required by the Nemo using \"ffmpeg\". if there is an option to give in m4a , let us know the script .Sharing the both audio here ([audio_data.zip](https://github.com/NVIDIA/NeMo/files/6141588/audio_data.zip)).\r\n\r\n2. Results using \"QuartzNet15x5Base-En\" model the transcript is pasted here.**\r\n\r\nO.\r\nNexi, okay.\r\nSo I am just reading an article related to deepplenartic architectures for action recognition.\r\nSo, Oh, this is Thi the article related to.\r\nExplore spacial temporal features.\r\nThrough various deep architectures, the role of optical flor is an import.\r\nThe impact ond real time capabilities and the compactness and indeve.\r\nRetability of the learn features.\r\nSo before?\r\nIn that, I'll just start with Er.\r\nSome paper or papers on the.\r\nYou see of fond, Zero, an dataset.\r\nOh.\r\nSo large scale video classification with convelsional red work.\r\nAnd two stream networks.\r\nDearis to popularity of single stream under two Streem network in action recondition.\r\nSo the single frame network is essentially an image classubition atwork.\r\nWith no temporal features.\r\nLate fusion uses too far apart frames and nfuses deep iancil features at the densely connected layers by flatktening.\r\nFeatures process from the two fribs.\r\nThe strength of the single stream strategieis that.\r\nWe can use transfer learning from model.\r\nTrained on last kilihis deterset.\r\nWe also have.\r\nNo one need to preporses the images for offical flow.\r\nEdirectly used the Rgbma Tata in this architecture.\r\nThis makes single stream network a candate for real time processing.\r\nIn the fusion architectures.\r\nDhid, the author proposed.\r\nThe number of cerometer increases significantly from a deep toti sanente.\r\nTo mitigate this, the author proposed.\r\nUsing a multiresorlation streng.\r\nOne, the dambets say high resolution for Viar steam on Ah, center crop of the video combined with Et.\r\nLow resolution contacts trim from the.\r\nHold vidio.\r\nI seen Inda.\r\nArchitecture, Iao.\r\nOky, so I'll just to or change the ear.\r\nContext.\r\nEr one secend ham, just ster.\r\nLet trile some.\r\nSo, generating optic, cof, floo, using, and videofloet.\r\nTo implementation.\r\nThe blog was originally published in Ta a blocg.\r\nBad slogan.\r\nIn a previous post, an introduction, tooptical flaws conducted, as well as an oervi of in its architecture, west of Pflonic.\r\nThis blot will focus in going deeper into optical floo.\r\nWhich will be done by generating optical Flo files?\r\nBoth from the standard.\r\nCintal Deta and the custom dance Vidio.\r\nIt will be conducted using a fork ofter.\r\nInvidioflown into pytorch cord bass, which can be found in the.\r\nFollowing link.\r\nOkay, I think I know a ond two three minutes. I talk.\r\nThank you.\r\nYama'am thank you.\r\nSo?\r\nYeai'll start for a.\r\nSome drenning stories.\r\nChecking.\r\nLike.\r\nDon't even O bac bag, okay?\r\nI'm selecting some.\r\nI'm sorry.\r\nA Ther deely, ie, no problem.\r\nAn ofa.\r\nYour background dis clear now.\r\nNothing, I'll go into an none.\r\nKi'm starting to read one article.\r\nMy speech is made up of many frequencies at the same time that it is a sum.\r\nOf all those frequencies.\r\nTo analyze the signal, we use the component frequencies as features.\r\nFor a transform is used to break the signal into.\r\nThese components.\r\nWe can use this splitting technique to convert the sound to a spectracram.\r\nGe frequency on the.\r\nMertical axis is splotted against.\r\nThe intensity of shady indicates the amplitude of the signal.\r\nTo create a spectrogram.\r\nThere are three lines sor.\r\nOf the cord.\r\nDivide the singer into.\r\nTime frequencies.\r\nSplit each frim.\r\nFrame singen into frequency components with an Ff. T.\r\nAnthardies is weat.\r\nEach timeframe is now represented with a vector of amplitude.\r\nAt each frequency.\r\nWel line up the vectors again.\r\nTheir time serius order, we have a.\r\nIual picture of the sound component.\r\nTheepectogram.\r\nNext you ill look at turfect.\r\nFeature exsection techniques which werd.\r\nWould reduce the noise and dimenslity of.\r\nOur data.\r\nSo?\r\nWhat his feature extruction with Ama Cisi.\r\nMetal frequency.\r\nCeptum cofision analysis is the deduction of an audio signal.\r\nTo essential speech component.\r\nPita.\r\nUsing both mel, frequency analysis and separate anasis.\r\nThe range of frequencies are reduced and bint.\r\nInto groups of frequencies that humans can distinguish.\r\nThe thing that is for ther.\r\nSeparated into sores and filter.\r\nSo that variations between speakers unrelated to.\r\nLatin can be filtered away.\r\nSo that is from my side.\r\nHank you?\r\nOcalre, the article whit have shared in the.\r\nThen I thought.\r\nYour tue.\r\nIn the social ignitic action isjed he minor of omanging a language. Anyone who speaks our language does so. in our accent, The way of the native speaker. speaker of the language speaks, Fo, languer, is define the standard ponudication, and is generally considered to be standard for Tesmantano.\r\nYour voice is not clar.\r\nNot too.\r\nCan use Um?\r\nYe fall nearer mouth.\r\nThe Mik never.\r\nEnglish has become an important language of O communication among the younger generations of the India, because of tic status as a language for the formal education, a large number of the young Indians in thebinokinnes.\r\nT thatif, they speak one of the Twentie Twola Indian language as their first language alongside eglists.\r\nAn implication of this is that when speaking English, the intervetion from phonology of their first language, enample, malalum gives rise to the accent in the speech of the Indian speakers of English.\r\nThis accent is generally very distinct and is deadly.\r\nIdentiable, for example, as the Malarlum, English, acpenthe, sengunlix, acenabungala bungari in English action derecto.\r\nThis younger genetration of Windia, is also a large, an a growing group of a user of a speech pace technology, togh a Hn.\r\nAndhelt device, and a voice assistant, This voice assistance has become very cool.\r\nEntifyigi English is spoken in the native accent, Altourh, nonative accent. Speece continues to challenge for them for Sa. Rc. System of the Voyce assists have knowledge that the speaker is going to speak with a certain accent. The voice assistante may be trimeg, to listen to Sudten feature in the voice, which would lead to a great performance security.\r\nBut sheud de not.\r\nO.\r\nYeah, ol kiv.\r\nOka.\r\nOh gesel, thank you.\r\nYen.\r\n\r\n**3. Results using \"QuartzNet15x5NR-En\" model the transcript is pasted here.**\r\nO Ol.\r\nNexsy O E.\r\nSo I am just reading an article related to Deplein Arctic architectures for action recognition.\r\nSo this is hi, the article related to.\r\nExplore spacial temporal features.\r\nThrough various deep architectures, the role of optical flour is an import.\r\nThe impact owned real time capabilities and the compactness a\r\nnd intevee.\r\nRetability of the learn features.\r\nSo before?\r\nIn that, I'll just start with her.\r\nSome paper or the papers on the.\r\nYou see a fone zero on at a set.\r\nSo lastscale video classification with convelsional red work.\r\nAnd two stream networks.\r\nEis to popularity of single stream, anda two Strepm network in action, a condition.\r\nSo the single frame network is essentially an image classivition atwork.\r\nWith no temporal features.\r\nAlate fusion uses two far apart frames and infuses deep irancill features at the densely connected layers by flatktning.\r\nFeatures processed from the two fribs.\r\nThe strength of the single stream strategy is ther.\r\nWe can use transfer learning from model.\r\nTrained on Lare killimi, etase.\r\nWe also have.\r\nNo, onnee to preposess the images for offical flow.\r\nDirectly used the Rgbmas Tata in this architectuce.\r\nThis makes single stream network a candate for real time processing.\r\nIn the fusion architectures.\r\nDid the author proposed?\r\nThe number of carameter increases significantly from a deep todi sanente.\r\nTo mitigate this, the author proposed.\r\nUsing a multiresolution street.\r\nN, the dambets a high resolution for vi steam on a center crop of the video combined with her.\r\nLow resolution contects trim from the.\r\nHold ridio.\r\nAs seen in the.\r\nArhitectui.\r\nOyso, I'll just tar or change the or.\r\nContext.\r\nErone, se cend am just stir.\r\nLet's change some.\r\nSogenerating optic co flow, usinga.\r\nTo implementation.\r\nThe blog was originally published in De a block.\r\nBad slogan.\r\nI aprevious post an introduction, Toptic Al Floos conducted, as well as an overview of Inix architecture, West offloic.\r\nThis blot will focus in going deeper into toptical flow.\r\nWhich will be done by generating optical flofiles?\r\nBoth from the standard.\r\nCentrel Deta and the custom dance ridio.\r\nIt will be conducted using he fork ofter.\r\nInvidio flow into pytoch hord bass, which can be found in the.\r\nFollowing link.\r\nOki think I know around to three minutes I talke.\r\nThank you.\r\nYama, thank you.\r\nSo?\r\nI'll start for E.\r\nSome reading stories.\r\nIcecking.\r\nEen o bu. o.\r\nI'm selecting some.\r\nI'am sorry.\r\nO.\r\nAno.\r\nBackground I ceno.\r\nNo, I'll go into an oe.\r\nAi'm starting to read one article.\r\nMy speech is made up of many frequencies at the same time thati it, isasum.\r\nAll those frequencies.\r\nTo analyze the signal, we use the component frequencies as features.\r\nFor a transform is used to break the signal into.\r\nThese components.\r\nWe can use this splitting technique to convert the sound to a spectrocram.\r\nGet frequency on the.\r\nVertical axess is plotted against.\r\nThe intensity of shady indicates the amplitude of the signal.\r\nTo create a spectrogram.\r\nThere are three lines her.\r\nEr.\r\nDivide the singar into.\r\nTime frequencies.\r\nSplit eas frim.\r\nFrame singgan into frequency components with an Ff. T.\r\nNtarti iseet.\r\nEach timeframe is now represented with a vector of amplitude.\r\nAt each frequency.\r\nWe line up the vectors again.\r\nIn theirtime series order, we have a.\r\nIsual picture of the sound component.\r\nThe expectocam.\r\nNext, you'll look at Teri.\r\nFeature exsection techniques picturded.\r\nWould reduce the noise and dimensility of.\r\nOur data.\r\nSo?\r\nWhat is feature extruction with Ama Piz?\r\nMetal frequency.\r\nSeptem cofision analysis is the deduction of an audio signal.\r\nTwo essential speech component.\r\nEt.\r\nUsing both mal, frequency analysis and seprate anlysis.\r\nThe range of frequencies are reduced and binn.\r\nInto groups of frequencies that humans can distinguish.\r\nThe thingtat is for Ther.\r\nSeparated into sores and filter.\r\nSo that variations between speakers unelated to.\r\nArticulation can be filtered away.\r\nSo that is from my side.\r\nHnyou.\r\nOkay, I'll read the article which has shared in the.\r\nHen, I thought.\r\nYe do.\r\nIn the social liitic actiont is wethe, minor ofper managingal language. Anyone who speaks our language does so in our accent, The way of the nativist speaker. speaker of the language speaks, o languageis, define ter standard pronuniciation and is generally considered to be standard or deferent action.\r\nYour voice is not cler.\r\nNot to goo.\r\nCan use em.\r\nYe fall near er mouth.\r\nEnglish has become an important language or for communication among the younger generation of the India, because of missus status as a language for the formal education, a large number of young Indians inarbinocene.\r\nThat if they speak one of the Twenti, two la Indian language as their first language alongside Iles.\r\nAn implication of this is that when speaking English, the intervesion from phonology of their first language, enample, Malalam gives srise to the accent in the speech of the Indian speakers of English.\r\nThis accent is generally very distinct and is deadly.\r\nIdentiable for the example, as the Madarlum English actenthor, sengunli, acentandbungala, bungai, in English action derectro.\r\nThis younger generation of Windia, is also a large and a growing group of a user of a steep pace technology through a han.\r\nAndhel device, and a Voyd assistant, Thiss woyce, The assistance has become very good.\r\nIentiying this English, S spoken in the lative accent, Olwer, nonative accents, Speece continues to challenge for them, for S. A Rc. system of the voyce assists have knowledge that the speaker is going to speak with a certain accent. The voice assistants may be trying to listen to sudden feature in the voice, which would lead to a great performance security.\r\nBut should be not.\r\nYe.\r\nYeah, I'l ges.\r\nOkin.\r\nOh, guesso, thank you.\r\nYen.\r\n", "Also provide some suitable or comparable tool /Script to record the audio using laptop /phone microphone, so that we get good transcript .It would be of great help, Thank you", "@jbalam-nv Can you please check on the sample audio and respective transcript and give us some inputs on how to improvise the result ?" ]

[]

[]

[ "thanks again @tiborkiss but we'll close this one for now because we can't pin torch version to the exact number like 1.7.1 We could add something like this right before upload to pypi though." ]

[ { "id": 1485815557, "node_id": "MDU6TGFiZWwxNDg1ODE1NTU3", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/bug", "name": "bug", "color": "d73a4a", "default": true, "description": "Something isn't working" } ]

[]

[ "PTL was recently fixed this problem here: https://github.com/PyTorchLightning/pytorch-lightning/commit/60d0c95e07b8edfa22f5892ebb72d93b1d35832f \r\nand they released `pytorch-lightening` 1.2.3 .\r\n\r\nTherefore I updated in `requirements/requirements.txt`\r\n```\r\n-pytorch-lightning>=1.1.0,<=1.1.5\r\n+pytorch-lightning>=1.2.3,<=1.2.3\r\n```\r\nThen I made a fresh new `venv` and installed NeMo from `r1.0.0rc1` with `./reinstall.sh`\r\n\r\nThat problem now related to `pytorch-lightening` version incompatibility with `torch` 1.8.0 has been solved.\r\nThen I start the first tutorial, in the following cell, now it crashes with another problem.\r\n\r\n```\r\nquartznet = nemo_asr.models.EncDecCTCModel.from_pretrained('QuartzNet15x5Base-En')\r\n[NeMo I 2021-03-09 22:19:02 cloud:56] Found existing object /home/tibor/.cache/torch/NeMo/NeMo_1.0.0rc1/QuartzNet15x5Base-En/00869f9c89b8393ca3de640e0c536bd2/QuartzNet15x5Base-En.nemo.\r\n[NeMo I 2021-03-09 22:19:02 cloud:62] Re-using file from: /home/tibor/.cache/torch/NeMo/NeMo_1.0.0rc1/QuartzNet15x5Base-En/00869f9c89b8393ca3de640e0c536bd2/QuartzNet15x5Base-En.nemo\r\n[NeMo I 2021-03-09 22:19:02 common:615] Instantiating model from pre-trained checkpoint\r\n---------------------------------------------------------------------------\r\nAttributeError Traceback (most recent call last)\r\n<ipython-input-6-38ceefde5424> in <module>\r\n----> 1 quartznet = nemo_asr.models.EncDecCTCModel.from_pretrained('QuartzNet15x5Base-En')\r\n\r\n~/asr/NeMo/nemo/core/classes/common.py in from_pretrained(cls, model_name, refresh_cache, override_config_path, map_location, strict, return_config)\r\n 616 if class_ is None:\r\n 617 class_ = cls\r\n--> 618 instance = class_.restore_from(\r\n 619 restore_path=nemo_model_file_in_cache,\r\n 620 override_config_path=override_config_path,\r\n\r\n~/asr/NeMo/nemo/core/classes/modelPT.py in restore_from(cls, restore_path, override_config_path, map_location, strict, return_config)\r\n 417 global _MODEL_RESTORE_PATH\r\n 418 _MODEL_RESTORE_PATH = os.path.abspath(os.path.expanduser(restore_path))\r\n--> 419 return cls._default_restore_from(restore_path, override_config_path, map_location, strict, return_config)\r\n 420 \r\n 421 @classmethod\r\n\r\n~/asr/NeMo/nemo/core/classes/modelPT.py in _default_restore_from(cls, restore_path, override_config_path, map_location, strict, return_config)\r\n 370 model_weights = path.join(tmpdir, _MODEL_WEIGHTS)\r\n 371 OmegaConf.set_struct(conf, True)\r\n--> 372 instance = cls.from_config_dict(config=conf)\r\n 373 instance = instance.to(map_location)\r\n 374 instance.load_state_dict(torch.load(model_weights, map_location=map_location), strict=strict)\r\n\r\n~/asr/NeMo/nemo/core/classes/common.py in from_config_dict(cls, config)\r\n 433 else:\r\n 434 # models are handled differently for now\r\n--> 435 instance = cls(cfg=config)\r\n 436 \r\n 437 if not hasattr(instance, '_cfg'):\r\n\r\n~/asr/NeMo/nemo/collections/asr/models/ctc_models.py in __init__(self, cfg, trainer)\r\n 87 def __init__(self, cfg: DictConfig, trainer: Trainer = None):\r\n 88 # Get global rank and total number of GPU workers for IterableDataset partitioning, if applicable\r\n---> 89 self.global_rank = 0\r\n 90 self.world_size = 1\r\n 91 self.local_rank = 0\r\n\r\n~/venv/pt8nemo/lib/python3.8/site-packages/torch/nn/modules/module.py in __setattr__(self, name, value)\r\n 993 buffers[name] = value\r\n 994 else:\r\n--> 995 object.__setattr__(self, name, value)\r\n 996 \r\n 997 def __delattr__(self, name):\r\n\r\nAttributeError: can't set attribute\r\n```\r\n\r\nThis looks like now the model is not compatible with `torch` 1.8.0.\r\n\r\nTherefore, I can still continue my work only if I apply https://github.com/NVIDIA/NeMo/pull/1869\r\nLooks like, it will take some time and effort to upgrade NeMo to use torch 1.8.0.", "A temporary workaround for myself it is here https://github.com/NVIDIA/NeMo/pull/1869", "@tiborkiss Please note that NeMo RC1 does not work on lightning 1.2.0. In fact, RC1 pins Lightning <= 1.1.5. If you need Lightning 1.2.0 or high, please use the NeMo main branch", "Thank you for the tips @blisc .\r\nLooks like the shortest workaround is to stick with torch 1.7.1", "In the setup.py we have a section related to `SUPPORTED_TORCH_VERSION` where is the `REMOVE AFTER 21.03 Container is released !` comment. \r\nAt least until this container is released it works, therefore I close this PR.", "@tiborkiss Sorry, what do you mean? Did you find a fix? Thanks!", "> @tiborkiss Sorry, what do you mean? Did you find a fix? Thanks!\r\n\r\nYes, in this branch, this file it does that fix. https://github.com/NVIDIA/NeMo/blob/r1.0.0rc1/setup.py\r\nAs soon as is not removed (read the comments there), it is enough. I just use this branch and it works." ]

[]

[]

[ "This is a pytorch lightning issue, please downgrade to 1.2.1 for the time being until they are able to support pytorch 1.8 as well as it's dependency of torchtext", "> This is a pytorch lightning issue, please downgrade to 1.2.1 for the time being until they are able to support pytorch 1.8 as well as it's dependency of torchtext\r\n\r\nMe neither :/ ", "I described already here https://github.com/NVIDIA/NeMo/issues/1868", "Duplicate of #1868" ]

[]

[]

[ "This pull request **introduces 4 alerts** when merging 03cb3ef0aad58dcaea2e849989897a94c849dcdc into 4d868b662af0fab87a4ad4078156486885bbb0a4 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-477438a4d8b220885f444f3455bf533d4efff55d)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "As per @MaximumEntropy suggestion, triangle mask can actually be dropped for inference (not just for export). I have verified that on a 100-sentence set (no diffs).", "This pull request **introduces 4 alerts** when merging 87d2435ef7a81a51cc3771ddda6231fad2d933b9 into 4d868b662af0fab87a4ad4078156486885bbb0a4 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-604b80f9af4f4b75ec5f1e8e4fddfba76c75e73a)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "This pull request **introduces 4 alerts** when merging d14de501f99f55b642cc61aa80c8b39ee1f99af2 into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-a0fadd69ffd0b4f84de7709fc606afd5480bf2c7)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "This pull request **introduces 4 alerts** when merging d934479f956f5db25dea8a1c5a38b4684deaa87f into e816905e286df545ce344e865ea6cf8e4b0c85ca - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-41c0615a46a7ec0c6eaf82a52f20acec4313b320)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "This pull request **introduces 4 alerts** when merging 9868414685036be59fa91740da4615a6ba24e3bb into 5d80627ca696384dc16e462f1e17efd33917a9dd - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-f2678947f893bee51552520cb19713b8d0e6632f)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "@borisfom unittests are failing", "> @borisfom unittests are failing\r\n\r\nWas there a change in CI machines setup? The tests still work on my machine, and I do not see changes in PT version in Dockerfile.\r\n================================================================================ slowest durations =================================================================================\r\n16.87s call tests/collections/nlp/test_nmt_model.py::TestMTEncDecModel::test_creation_saving_restoring\r\n10.79s call tests/collections/nlp/test_nmt_model.py::TestMTEncDecModel::test_cpu_export\r\n10.51s call tests/collections/nlp/test_nmt_model.py::TestMTEncDecModel::test_gpu_export\r\n\r\n(6 durations < 0.005s hidden. Use -vv to show these durations.)\r\n========================================================================= 3 passed, 20 warnings in 42.35s ==========================================================================\r\nboris@snikolaev-DGXStation:~/git/NeMo$ sudo pip3 list | grep torch\r\npytorch-lightning 1.2.2\r\ntorch 1.7.1+cu110\r\ntorch-stft 0.1.4\r\ntorchaudio 0.7.2\r\ntorchtext 0.8.0\r\ntorchvision 0.8.2+cu110\r\ntrtorch 0.2.0a0\r\n", "@borisfom can you try inside the container? main branch uses 21.02-py3 for CI", "This pull request **introduces 4 alerts** when merging 162a4b41dde512b057aad5f990c2344bce2908c1 into 05eb2587a95e512b5e95705d9f77582634abbd95 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-ac780a854af2a49debda5da204bee3d09754fd53)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "This pull request **introduces 4 alerts** when merging feb2aa98a412a0522af824985c077d0ea6380b94 into 05eb2587a95e512b5e95705d9f77582634abbd95 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-91c724a41e11640e779d8bba07b580a716b769b0)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "> @borisfom can you try inside the container? main branch uses 21.02-py3 for CI\r\n\r\nRight, so I did - also bumped the base in Dockerfile and enabled Megatron export test that works with 21.02", "This pull request **introduces 4 alerts** when merging c9c2978935a5618d4452bd69b00c9db35658ada2 into 646aff11aa788b7d874dae930ce6ffcd658c1f16 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-18195fa22e39d5a4c0088a69831cfed2ce88b4a8)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import", "> > @borisfom can you try inside the container? main branch uses 21.02-py3 for CI\r\n> \r\n> Right, so I did - also bumped the base in Dockerfile and enabled Megatron export test that works with 21.02\r\n\r\nAfter fixing Torch compatibility issues with upgraded PT, the CI fails on unrelated test:\r\ncd tools/ctc_segmentation && pip install -r requirements.txt && DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends ffmpeg\r\ntest -f /etc/shinit_v2\r\n+ source /etc/shinit_v2\r\n+++ sed -n 's/^NVRM.*Kernel Module *\\([^() ]*\\).*$/\\1/p' /proc/driver/nvidia/version\r\n+++ sed 's/^$/unknown/'\r\n++ NV_DRIVER_VERS=460.39\r\n++ export _CUDA_COMPAT_PATH=/usr/local/cuda/compat\r\n++ _CUDA_COMPAT_PATH=/usr/local/cuda/compat\r\n+++ hostname\r\n++ _CUDA_COMPAT_CHECKFILE=/usr/local/cuda/compat/.460.39.7fb21172b5cf.checked\r\n++ _CUDA_COMPAT_REALLIB=/usr/local/cuda/compat/lib.real\r\n++ _CUDA_COMPAT_SYMLINK=/usr/local/cuda/compat/lib\r\n++ '[' -n 460.39 -a -e /dev/nvidiactl -a '!' -e /usr/local/cuda/compat/.460.39.7fb21172b5cf.checked ']'\r\n+++ cat /sys/module/mlx5_core/version\r\n+++ true\r\n++ DETECTED_MOFED=\r\n++ dpkg --compare-versions 0 lt 4.9\r\n++ export 'OMPI_MCA_pml=^ucx'\r\n++ OMPI_MCA_pml='^ucx'\r\n++ unset _CUDA_COMPAT_CHECKFILE\r\n++ unset _CUDA_COMPAT_REALLIB\r\n++ unset _CUDA_COMPAT_SYMLINK\r\n+ '[' -z '' ']'\r\n+ return\r\n+ cd tools/ctc_segmentation\r\n+ pip install -r requirements.txt\r\nCollecting ctc_segmentation==1.1.0\r\n Downloading ctc_segmentation-1.1.0.tar.gz (57 kB)\r\nCollecting num2words\r\n Downloading num2words-0.5.10-py3-none-any.whl (101 kB)\r\nRequirement already satisfied: setuptools in /opt/conda/lib/python3.8/site-packages (from ctc_segmentation==1.1.0->-r requirements.txt (line 1)) (50.3.1.post20201107)\r\nRequirement already satisfied: numpy in /opt/conda/lib/python3.8/site-packages (from ctc_segmentation==1.1.0->-r requirements.txt (line 1)) (1.19.2)\r\nRequirement already satisfied: Cython in /opt/conda/lib/python3.8/site-packages (from ctc_segmentation==1.1.0->-r requirements.txt (line 1)) (0.28.4)\r\nCollecting docopt>=0.6.2\r\n Downloading docopt-0.6.2.tar.gz (25 kB)\r\nBuilding wheels for collected packages: ctc-segmentation, docopt\r\n Building wheel for ctc-segmentation (setup.py): started\r\n Building wheel for ctc-segmentation (setup.py): finished with status 'done'\r\n Created wheel for ctc-segmentation: filename=ctc_segmentation-1.1.0-cp38-cp38-linux_x86_64.whl size=120481 sha256=6583c5221f311b58278d169b5ad53ba162917094963459610a404440fb6d6759\r\n Stored in directory: /root/.cache/pip/wheels/4e/6e/9e/2904d4b4f46b220690572ef09688e6cf248a6340cd2758742c\r\n Building wheel for docopt (setup.py): started\r\n Building wheel for docopt (setup.py): finished with status 'done'\r\n Created wheel for docopt: filename=docopt-0.6.2-py2.py3-none-any.whl size=13704 sha256=09cec6b7a65ec64f9ee6bb37422283e3ed48fa51d054016adb6da1201a39e327\r\n Stored in directory: /root/.cache/pip/wheels/56/ea/58/ead137b087d9e326852a851351d1debf4ada529b6ac0ec4e8c\r\nSuccessfully built ctc-segmentation docopt\r\nInstalling collected packages: ctc-segmentation, docopt, num2words\r\nSuccessfully installed ctc-segmentation-1.1.0 docopt-0.6.2 num2words-0.5.10\r\n+ DEBIAN_FRONTEND=noninteractive\r\n+ apt-get install -y --no-install-recommends ffmpeg\r\nReading package lists...\r\nBuilding dependency tree...\r\nReading state information...\r\nE: Unable to locate package ffmpeg\r\nscript returned exit code 100\r\nClearly CI issue - ffmpeg os one of the pre-reqs to install NeMo, mentoned both in README and Dockerfile:\r\nUse this installation mode if you want the latest released version.\r\n.. code-block:: bash\r\n apt-get update && apt-get install -y libsndfile1 ffmpeg\r\n pip install Cython\r\n pip install nemo_toolkit[all]==1.0.0b3\r\n", "This pull request **introduces 4 alerts** when merging 874f1f65c64abf867643f8f7094cfd1ad32f99c9 into b0f9f94469b1b7bacb10fb1851f8fabfd8b4323d - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-fbdba08fb50cdc68c493f8dca8e7c382bbc5a95a)\n\n**new alerts:**\n\n* 2 for Modification of dictionary returned by locals\\(\\)\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "buncha tests are failing but i dont think theyre related to this pr " ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ "Commits not signed." ]

[]

[]

[]

[]

[]

[ "thanks!" ]

[]

[]

[ "@okuchaiev Maybe there is some information on this issue? Thanks!", "There are several notebooks for transfer learning char and subword models in Nemo now, please have a look and reopen the issue if they are insufficient. " ]

[]

[]

[ "This pull request **introduces 1 alert** when merging 9de7bc589925be7b77ffcaa8fad5ffe14115971e into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-9075874939d4e184867f5ce3384215d359cffa1d)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 8f8a275bc0321594d826a55b91750ea1cfd7b309 into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-5702410f047f78f5ebaca2b3ad87b84a369204ef)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging e454c4c70dee1a67c73d14585861fc4610532a80 into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-165135c81c13b53411b41f1e0f9f5cf9f515bee4)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 4946314772b952f444e1987916b49a47f1f23534 into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-b0d15459dd8c2abe527d715a5a24ac92a14546ae)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging ea51c3bc89ae0e1f82c716660d23292975adac0f into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-5e16827ec1a959c6e51f28f101ab929799a0b6db)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging aea2b7278270f0d796712b3a079b390d9c718885 into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-3353ae343a0372e2b188232c07455155bf4163d7)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging bcb683ae9433d3c7c3ccfae71f648932e2d8249d into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-abc49ad19b712e818b0805472983308351fb9d74)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 5addd5014507e769fdc2700e14e501183bf54bd7 into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-1c0a2b37f5c7205bf9a1c9662429605a9df6e7e3)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging af028e9e1e7a3a40389c59790868c181cab9607d into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-45592136e0f1e62132408bb0de4eedcb314936d9)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 8edd7f134be7cf2b2d415cb3123bd0a2120e13cb into 74a1ed8618872d331fcea3133d47b8b3552c435f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-b4c1efb631689292d9cf6c19c8a2a488b511fea1)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 5bf8f3dd8229d94784d1f301d9ab5987a70e5813 into e8912ca6e3321347272a6a7da18e052812fb2062 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-646fe486329bfa5221e90de446743d60ef1dc076)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging b3f93aea208c0ab36b565b512e82917ffa6b7339 into aaab31781309fa2c6a3e5e3cb91e4ee7f4d002c8 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-b0c16830e674bdfaf523e442d9a41a66f39909df)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 9cdc539c6fa572a78284a4f9dd2fdccf25b16f2d into d623109ee752f654355c141bae43b64b2b98dba8 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-6474142b94e4abb4a8b7d85403bbbbbf08fc112a)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 19215a16d84f72539a08ce91f99d8e1aa64dc33e into 480bbc39b8c35f2f5ff0e2a2b61f4e2526de28eb - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-621b2b8007367a6fb5be4448bf9d3840f1ae809e)\n\n**new alerts:**\n\n* 1 for Unused local variable" ]

[]

[]

[ "@okuchaiev done" ]

[]

[ { "login": "ericharper", "id": 11999610, "node_id": "MDQ6VXNlcjExOTk5NjEw", "avatar_url": "https://avatars.githubusercontent.com/u/11999610?v=4", "gravatar_id": "", "url": "https://api.github.com/users/ericharper", "html_url": "https://github.com/ericharper", "followers_url": "https://api.github.com/users/ericharper/followers", "following_url": "https://api.github.com/users/ericharper/following{/other_user}", "gists_url": "https://api.github.com/users/ericharper/gists{/gist_id}", "starred_url": "https://api.github.com/users/ericharper/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/ericharper/subscriptions", "organizations_url": "https://api.github.com/users/ericharper/orgs", "repos_url": "https://api.github.com/users/ericharper/repos", "events_url": "https://api.github.com/users/ericharper/events{/privacy}", "received_events_url": "https://api.github.com/users/ericharper/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "Hi, restoring NLP models can be complicated as there are potentially many different artifacts needed for restore. This is why we've created the .nemo file and .restore_from. We mainly use the PyTorch Lightning Checkpoints to auto-resume from training. They can be used in other workflows outside of NeMo, but then the user would need to add custom implementation for restore.\r\n\r\nAnd to answer your question \"I did not know where did the megatron-bert-cased_encoder_config.json come from?\"\r\nMegatron models need certain artifacts to be instantiated, mainly the config and vocab. We generate both of these and automatically save them in the .nemo file when using Megatron models in NeMo NLP.\r\n\r\nAlso, we will be updating the NeMo NLP documentation with this information in the near future. Sorry for the current inconvenience." ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ "Suggestion to change the folder name from speaker_diarization to speaker_recognition in scripts.\r\nI couldn't comment on unchanged files. Some links in speaker recognition notebooks point to these scripts. \r\nSome of them:\r\n\r\n> in tutorials/Speaker_Recognition_Verification.ipynb\r\n\r\n* get_hi-mia-data.py (https://github.com/NVIDIA/NeMo/blob/rir_noise_aug_2/scripts/get_hi-mia_data.py) script to download the necessary files -> get_hi-mia-data.py (https://github.com/NVIDIA/NeMo/blob/main/scripts/speaker_recognition/get_hi-mia_data.py) script to download the necessary files\r\n* wget -P scripts https://raw.githubusercontent.com/NVIDIA/NeMo/$BRANCH/scripts/scp_to_manifest.py -> wget -P scripts https://raw.githubusercontent.com/NVIDIA/NeMo/$BRANCH/scripts/speaker_recogntion/scp_to_manifest.py", "This pull request **introduces 4 alerts** when merging 26e75e512ebd029be9c09f92a61ca2ad7d0fb257 into 89a8ea037a121001e105e95b49d71906a8de29aa - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-18e7412bcf8ad035dda398a0b8f5fffef1c6d2db)\n\n**new alerts:**\n\n* 4 for Unused import", "I've checked scripts and tutorials for speech command, VAD and speaker diarization. \r\n\r\nCould you please change\r\n1. **NeMo/scripts/** in examples/asr/vad_infer to **NeMo/scripts/voice_activity_detection**\r\n2. multiple **<NEMO_ROOT>/scripts** in examples/asr/speech_to_label.py to correposinding subfolders.\r\n3. in tutorials/asr/06 VAD: scripts/voice_activity_recognition/process_vad_data -> scripts/**dataset_processing**/process_vad_data\r\n4. in tutorials/speaker_recognition/Speaker_Diarization_Inference.ipynb: \"you can also tune the threshold on your dev set. Use this provided script\" fix the link of the **script**. It would be much appreciated if you can also help change the link to https://github.com/NVIDIA/NeMo/blob/r1.0.0rc1/nemo/collections/asr/parts/vad_utils.py in the above line. \r\n\r\nThanks!\r\n", "> \r\n> \r\n> I've checked scripts and tutorials for speech command, VAD and speaker diarization.\r\n> \r\n> Could you please change\r\n> \r\n> 1. **NeMo/scripts/** in examples/asr/vad_infer to **NeMo/scripts/voice_activity_detection**\r\n> \r\n> 2. multiple **<NEMO_ROOT>/scripts** in examples/asr/speech_to_label.py to correposinding subfolders.\r\n> \r\n> 3. in tutorials/asr/06 VAD: scripts/voice_activity_recognition/process_vad_data -> scripts/**dataset_processing**/process_vad_data\r\n> \r\n> 4. in tutorials/speaker_recognition/Speaker_Diarization_Inference.ipynb: \"you can also tune the threshold on your dev set. Use this provided script\" fix the link of the **script**. It would be much appreciated if you can also help change the link to https://github.com/NVIDIA/NeMo/blob/r1.0.0rc1/nemo/collections/asr/parts/vad_utils.py in the above line.\r\n> \r\n> \r\n> Thanks!\r\n\r\nThanks for checking them out. I missed them.\r\n1-done\r\n2-done\r\n3-done\r\n4-Please check it again to make sure I put the right links. We need to change them to main after the release?", "> > I've checked scripts and tutorials for speech command, VAD and speaker diarization.\r\n> > Could you please change\r\n> > ```\r\n> > 1. **NeMo/scripts/** in examples/asr/vad_infer to **NeMo/scripts/voice_activity_detection**\r\n> > \r\n> > 2. multiple **<NEMO_ROOT>/scripts** in examples/asr/speech_to_label.py to correposinding subfolders.\r\n> > \r\n> > 3. in tutorials/asr/06 VAD: scripts/voice_activity_recognition/process_vad_data -> scripts/**dataset_processing**/process_vad_data\r\n> > \r\n> > 4. in tutorials/speaker_recognition/Speaker_Diarization_Inference.ipynb: \"you can also tune the threshold on your dev set. Use this provided script\" fix the link of the **script**. It would be much appreciated if you can also help change the link to https://github.com/NVIDIA/NeMo/blob/r1.0.0rc1/nemo/collections/asr/parts/vad_utils.py in the above line.\r\n> > ```\r\n> > \r\n> > \r\n> > Thanks!\r\n> \r\n> Thanks for checking them out. I missed them.\r\n> 1-done\r\n> 2-done\r\n> 3-done\r\n> 4-Please check it again to make sure I put the right links. We need to change them to main after the release?\r\n\r\nThe **here** one should be https://github.com/NVIDIA/NeMo/blob/r1.0.0rc1/nemo/collections/asr/parts/vad_utils.py\r\nThe **script** one should be https://github.com/NVIDIA/NeMo/blob/r1.0.0rc1/scripts/voice_activity_detection/vad_tune_threshold.py\r\n\r\nYes after the release. \r\nThanks.", "I see you are merging to rc1 branch but links are pointed to main. If we test before merging to main, notebook tutorials gonna fail ", "> \r\n> \r\n> I see you are merging to rc1 branch but links are pointed to main. If we test before merging to main, notebook tutorials gonna fail\r\n\r\nIt looks tricky. I tried not to change those parts. The code should work with both actually as RC is merged into main everyday. So if it is based on main, it was going to get tested with main anyway and this PR is not going to affect the results. Should I change all to main or RC?", "> > I see you are merging to rc1 branch but links are pointed to main. If we test before merging to main, notebook tutorials gonna fail\r\n> \r\n> It looks tricky. I tried not to change those parts. The code should work with both actually as RC is merged into main everyday. So if it is based on main, it was going to get tested with main anyway and this PR is not going to affect the results. Should I change all to main or RC?\r\n\r\nActually I just see [one notebook](https://github.com/VahidooX/NeMo/blob/organize_scripts/tutorials/speaker_recognition/Speaker_Recognition_Verification.ipynb) script (get_hi-mia-data.py) to main branch, but rest of notebooks looks good. Also its path should be changed to `dataset_processing` not `speaker_recognition`. Just this rest all looks good to me. I vote main since for later releases we don't need to change again. ", "> \r\n> \r\n> > > I see you are merging to rc1 branch but links are pointed to main. If we test before merging to main, notebook tutorials gonna fail\r\n> > \r\n> > \r\n> > It looks tricky. I tried not to change those parts. The code should work with both actually as RC is merged into main everyday. So if it is based on main, it was going to get tested with main anyway and this PR is not going to affect the results. Should I change all to main or RC?\r\n> \r\n> Actually I just see [one notebook](https://github.com/VahidooX/NeMo/blob/organize_scripts/tutorials/speaker_recognition/Speaker_Recognition_Verification.ipynb) script (get_hi-mia-data.py) to main branch, but rest of notebooks looks good. Also its path should be changed to `dataset_processing` not `speaker_recognition`. Just this rest all looks good to me. I vote main since for later releases we don't need to change again.\r\n\r\nThanks for catching them! Updated those. I also prefer main, but for now, I made them all RC as we are need to change others to master later anyway. " ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "Thanks @ekmb , just did :) ", "yes please, this looks like bugfix for r1.0.0rc1", "Thanks @ekmb and @okuchaiev - rebasing to r1.0.0rc1 resulted in many file changes, but notebook hasn't changed at all.", "This pull request **introduces 4 alerts** when merging f6e7fc182f103e0b58c47ccb39645995ac178324 into bc38019ffa144dcf0bfc735a9838bb0e6fba71ab - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2d6425e234837de89ff7012c4a2fb38315c3ad09)\n\n**new alerts:**\n\n* 4 for Unused import" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]