Datasets:

---

renwei2024

/

nemo-github-issues

like 0

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ "No, that flag can accept a list of GPU ids. https://pytorch-lightning.readthedocs.io/en/stable/trainer.html#gpus \r\nSo that placeholder looks right to me." ]

[ { "id": 4910680898, "node_id": "LA_kwDOC_bI7s8AAAABJLMLQg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/TN/ITN", "name": "TN/ITN", "color": "0052cc", "default": false, "description": "" } ]

[ { "login": "yzhang123", "id": 4204271, "node_id": "MDQ6VXNlcjQyMDQyNzE=", "avatar_url": "https://avatars.githubusercontent.com/u/4204271?v=4", "gravatar_id": "", "url": "https://api.github.com/users/yzhang123", "html_url": "https://github.com/yzhang123", "followers_url": "https://api.github.com/users/yzhang123/followers", "following_url": "https://api.github.com/users/yzhang123/following{/other_user}", "gists_url": "https://api.github.com/users/yzhang123/gists{/gist_id}", "starred_url": "https://api.github.com/users/yzhang123/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/yzhang123/subscriptions", "organizations_url": "https://api.github.com/users/yzhang123/orgs", "repos_url": "https://api.github.com/users/yzhang123/repos", "events_url": "https://api.github.com/users/yzhang123/events{/privacy}", "received_events_url": "https://api.github.com/users/yzhang123/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "No. Currently NeMo ASR models transcribe 55 as \"fifty five\". \r\nIn fact, if En normalization is tuned on (default) in data layer, then 55 will be converted to fifty five.\r\n\r\nKeep in mind that there could be many valid ways to say, for example, 2020 differently - you can say \"two thousands twenty\" or \"twenty twenty\". ", "Hi @okuchaiev thanks for your response. I was thinking that this could be an NLP task rather than expecting ASR to generate such results similar to punctuation and capitalization model where it do some post-processing to the enhance the ASR results further more.", "Found a related\r\npaper: https://isca-speech.org/archive/Interspeech_2017/pdfs/1274.PDF\r\nblogpost: https://machinelearning.apple.com/research/inverse-text-normal\r\n\r\nbut there seems to be no code/pretrained models present for this.", "@rohitgr7 Stay tuned, we are working on this :)\r\nIn the meantime, we have released a tool for text normalization, check it out here: https://colab.research.google.com/github/NVIDIA/NeMo/blob/main/tutorials/tools/Text_Normalization_Tutorial.ipynb", "@yzhang123 Awesome!! eagerly waiting for this :smiley: ", "@rohitgr7 check out this PR https://github.com/NVIDIA/NeMo/pull/1797 :)\r\nPlease reopen this issue if you have followup questions", "thanks, @yzhang123. Will check it out :)", "@rohitgr7 and let us know your feedback, or if anything does not work :)\r\n\r\n" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "are these fixes to 1.0.0b2? Why is this sent to main?\r\n" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "This pull request **introduces 1 alert** when merging 9c18170f1ed653b80bdf62dc30c810dbf3e5b87f into 48620bb1369dc72dcac7b9a4e75b113f1cc55d36 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-29b3095947becb5ca10111ca89fa4e93061b71fb)\n\n**new alerts:**\n\n* 1 for Unused import", "is this a fix? should we send this to v1.0.0b2 as well?", "> is this a fix? should we send this to v1.0.0b2 as well?\r\n\r\nboth v1.0.0b2 and main are Ok, it's not a bug just removing these to avoid confusion for users that any model from get_lm() works with all NLP tasks.", "changed base to v1.0.0b2\r\n" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "@SeanNaren when you commit, could you please do \"git commit -s ...\" to \"sign\" your commits.\r\nFor the already committed commits, DCO's bot output tells you https://github.com/NVIDIA/NeMo/pull/1361/checks?check_run_id=1321341664 how to sign already committed commits. \r\n\r\nThe -s flag means: \"certifies that committer has the rights to submit this work under the same license and agrees to a Developer Certificate of Origin (see http://developercertificate.org/ for more information)\". We thought that asking for this is an easier ask then asking every external contributor to sign some sort of CLA.\r\n", "Will do! I think I made a mess with the DCO commands, will fix...", "@okuchaiev should all be done!" ]

[]

[]

[ "Reformatted ipynb using Colab as per Vahid's recommendation" ]

[]

[]

[ "Sadly, I don't know any right-to-left language so I can not comment on specifics.\r\n\r\nHowever, I would imagine that right-to-left could pose a problem for CTC mechanism because it tries to learn alignments left-to-right. Have you tried doing a mirror reflection of your transcripts (e.g. converting them to left-to-right), then trying to learn such model first. Then, if it works, you can just mirror predictions and obtain the proper spelling?\r\n\r\nAlso, learning rate of 0.000001 is terribly small. Are you using Novograd? Try 0.001 maybe higher, maybe lower - this is something that you need to experiment with." ]

[]

[]

[]

[]

[]

[]

[]

[ { "login": "blisc", "id": 4763352, "node_id": "MDQ6VXNlcjQ3NjMzNTI=", "avatar_url": "https://avatars.githubusercontent.com/u/4763352?v=4", "gravatar_id": "", "url": "https://api.github.com/users/blisc", "html_url": "https://github.com/blisc", "followers_url": "https://api.github.com/users/blisc/followers", "following_url": "https://api.github.com/users/blisc/following{/other_user}", "gists_url": "https://api.github.com/users/blisc/gists{/gist_id}", "starred_url": "https://api.github.com/users/blisc/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/blisc/subscriptions", "organizations_url": "https://api.github.com/users/blisc/orgs", "repos_url": "https://api.github.com/users/blisc/repos", "events_url": "https://api.github.com/users/blisc/events{/privacy}", "received_events_url": "https://api.github.com/users/blisc/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "Just To let you know that I trained Tacotron2 without wandb logging and it worked successfully ", "That makes sense, since I run TTS experiments with the tensorboard logger only and add_image() is tensorboard's API. I'll add a check so it only logs to tensorboard. If you want to log images to WandB, feel free to submit a PR with that change", "> That makes sense, since I run TTS experiments with the tensorboard logger only and add_image() is tensorboard's API. I'll add a check so it only logs to tensorboard. If you want to log images to WandB, feel free to submit a PR with that change\r\n\r\n@blisc could tell me please how to disable logging images into WandB? I simply want to track the experiment with WandB that's all " ]

[]

[]

[]

[ { "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" } ]

[ { "login": "vsl9", "id": 4344862, "node_id": "MDQ6VXNlcjQzNDQ4NjI=", "avatar_url": "https://avatars.githubusercontent.com/u/4344862?v=4", "gravatar_id": "", "url": "https://api.github.com/users/vsl9", "html_url": "https://github.com/vsl9", "followers_url": "https://api.github.com/users/vsl9/followers", "following_url": "https://api.github.com/users/vsl9/following{/other_user}", "gists_url": "https://api.github.com/users/vsl9/gists{/gist_id}", "starred_url": "https://api.github.com/users/vsl9/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/vsl9/subscriptions", "organizations_url": "https://api.github.com/users/vsl9/orgs", "repos_url": "https://api.github.com/users/vsl9/repos", "events_url": "https://api.github.com/users/vsl9/events{/privacy}", "received_events_url": "https://api.github.com/users/vsl9/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "Changing [here](https://github.com/NVIDIA/NeMo/blob/6796faa62e9d9c42ee86f860f1a79d8db15561ae/nemo/collections/asr/modules/beam_search_decoder.py#L57) `return {\"predictions\": NeuralType(('B', 'T'), PredictionsType())}` to `return {\"predictions\": [NeuralType(('B', 'T'), PredictionsType())]}` (note the square brackets) also helps due to [this](https://github.com/NVIDIA/NeMo/blob/6796faa62e9d9c42ee86f860f1a79d8db15561ae/nemo/core/classes/common.py#L285) check, as the comment \"the provided NeuralTypes {} should enable container support (add '[]' to the NeuralType definition)\" [here](https://github.com/NVIDIA/NeMo/blob/6796faa62e9d9c42ee86f860f1a79d8db15561ae/nemo/core/classes/common.py#L296) suggests. What is a correct way to fix this output type definition? " ]

[]

[]

[ "Thank you!" ]

[]

[]

[ "You can use .save_to/.restore_from methods to save to and restore from \".nemo\" file. \r\n\".nemo\" file is a simply \".tar.gz\" file (you can rename it to tar.gz and unpack) with checkpoint weights, configuration yaml which was used and (for some models) additional artifacts such as tokenizer's vocabulary, etc. " ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ "This pull request **introduces 5 alerts** when merging 8c0e7e3d211117ebadddc45c3491357adb7fc085 into 94b64d32c0c54fd680b34c3b7dfb664b1035481b - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-590b9ded82eb997004669a408fbab57b8f746582)\n\n**new alerts:**\n\n* 3 for Unused import\n* 1 for Inconsistent method resolution order\n* 1 for Unused local variable", "I have started the review. It is 2k+ added lines of code and may take some time. In the meanwhile:\r\n+ Please sign off all your commits. Your previous commits should also get signed. The DCO check gives you the commands how to do that.\r\n+Run \"python setup.py style --fix\" in the main folder to autoformat the style of the code.\r\n+ There are some LGTM warnings to get fixed.\r\n+ You may pull from main to make it up to date as some important changes have happened recently", "It's definitely far from done, and I think I've done a lot of changes here that are not required per say for wav2vec! Hoping to get back to working on this soon", "> \r\n> \r\n> It's definitely far from done, and I think I've done a lot of changes here that are not required per say for wav2vec! Hoping to get back to working on this soon\r\n\r\nIt is a large PR with more than 2500 lines. It may take some time, and may need more than one round of review. I did a quick review and put some comments. You may take a look at them whenever you can. We can do another round of review when it is closer to be working.", "This pull request **introduces 5 alerts** when merging 5adcca5f702ff612251577ae937da07945af4a17 into 75774cf24755bddb393e3b1a4cbcc1921daf8898 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-d48d57930ccb41cdffb6ffbd9927de928ef56599)\n\n**new alerts:**\n\n* 3 for Unused import\n* 1 for Inconsistent method resolution order\n* 1 for Unused local variable", "This pull request **introduces 3 alerts** when merging 859fb90e01c55d15d1b8fba31ef213df95a4fca4 into 75774cf24755bddb393e3b1a4cbcc1921daf8898 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-9218028773ddc1cb058ee252fc4846b02855376e)\n\n**new alerts:**\n\n* 1 for Inconsistent method resolution order\n* 1 for Unused local variable\n* 1 for Unused import", "This pull request **introduces 3 alerts** when merging 6ea68ca7fae1bfd212f2286d9f590b06f913c6b8 into ece60b02f10494d73b3ca109d2c4fa6ef556ec47 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2fe1f316ff31b2a3d8065a960a1068e0cea14c89)\n\n**new alerts:**\n\n* 1 for Inconsistent method resolution order\n* 1 for Unused local variable\n* 1 for Unused import", "This pull request **introduces 3 alerts** when merging 9281c2753e744be773c3d15ffc60b267a3e30b03 into f0eeaed82d975b7ba1833185eadda541d004de51 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-ef8a924d9ae72d209dc564e4fb6f89409c1071e4)\n\n**new alerts:**\n\n* 1 for Inconsistent method resolution order\n* 1 for Unused local variable\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging ad6f9606cd4cc3386b00b4b83cfe88ecd7c9c17b into 99ed9c2ae5c840a8723f236530ccb1abe8c75377 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-885ae99a66fdfdadbb3c1c2e7f4201bb133fc412)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging e0609f220b7892c48e57d7c15381e4268bb8be8e into 99ed9c2ae5c840a8723f236530ccb1abe8c75377 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-79e6baa6a67092c16f519462fc92217a9be1eb04)\n\n**new alerts:**\n\n* 1 for Unused local variable", "I'm still plugging away at this! Currently working on fixing the _asr part of the model.", "This pull request **introduces 1 alert** when merging 6b222801860b53af3f53b56602de97b7f1de251d into 5b6dd3850129898be47cf0d65587897ec45a5b59 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2e7b9ab2f75863578795c7e1095dd415e1c864d8)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 1 alert** when merging 931f52012a65adc1110b309493fc3bfd8467d2e7 into 5b6dd3850129898be47cf0d65587897ec45a5b59 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-d5d0615c7db662d2f81b690aa617611d5cbed223)\n\n**new alerts:**\n\n* 1 for Unused local variable", "This pull request **introduces 2 alerts** when merging c796af5463e19dc909943dd2e76425271088fd25 into 5b6dd3850129898be47cf0d65587897ec45a5b59 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-0a26c3076a69e3d2bc054f0989cf5b1b190ddf9f)\n\n**new alerts:**\n\n* 1 for Unused local variable\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging e208c1a3be7b85bf82c6fc7abe1bc4c955e8e360 into 5b6dd3850129898be47cf0d65587897ec45a5b59 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-c2732d6b47b826a8cf7ff9e7dbd5ab0b68f6ce93)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 9e4ba49392bc80ed5d807c453e776da5d38b0c87 into 5b6dd3850129898be47cf0d65587897ec45a5b59 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-f28d04442be47dc9858ee62568f2c877d0e1dce2)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 2 alerts** when merging 07eb404e25b6f05dfcd15e251164c10f10b221df into 30b2ed2a84b5e568d15b97079b15a9559450f480 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-7873b032551df5e45b13b07c8a75437785b5bf93)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging f95b247d75694874b75562f815436de179792c54 into 4e804fd88bc197d031e97972957d8ad21e0c002f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-90f03bcc7d66c9fd1c19abd5688ee1cab11f8fbf)\n\n**new alerts:**\n\n* 2 for Unused import", "Thanks for all the suggestion @VahidooX @okuchaiev, I addressed most of them but need to follow up on a few, let me give you guys a status update:\r\n\r\nWhats been done:\r\n\r\n- The code works! I've successfully pre-trained a model on AN4/fine-tuned on AN4, and am working on testing on mozilla common voice right now. Novograd seems to work well, so will be using that instead of adam. I don't have much GPU resources (using AWS atm) but will leave it over the weekend and see how pre-training goes.\r\n- I've cleaned up the code extensively. A lot of the fairseq code seemed very weirdly interconnected so I spent a lot of time ensuring the code flow makes sense, and adding a bit of explanation where necessary. I've removed a lot of modules which were not necessary with latest AMP changes and such, and rely on torch modules where I can.\r\n- Code sits on all existing NeMo code now with no code changes required :)\r\n\r\nWhat needs to be addressed:\r\n\r\n- (EDIT: done)! Padding mask is being generated in the forward pass of the wav2vec model. I tested the runtime here and it's very minimal overhead but not vectorized. I'm going to try vectorize it.\r\n- I created wav2vec base to wrap basic dataloader functions that I copied from the ctc_model.py. I feel like there is a refactor that can happen to make this an inheritance. I assume NeMo never ran into this because the ctc_model class worked well for all models supported\r\n- As suggested, pull out the fine-tuning portion for a separate PR. But I'll keep it here for the time-being for model training sake as we ensure the codebase works. \r\n\r\nMy current steps:\r\n\r\n- Train mozilla pre-trained model, ensure convergence. Ensure fine-tuning works, and compare against a weaker baseline to ensure the model has learnt something from pre-training. EDIT: I'm running into OOM issues, if the model can be trained on a 32gb or higher VRAM card this will be beneficial. Currently I'm pruning the manifest to 7 second files to keep the memory low on a 16gb VRAM card.\r\n- Pull out fine-tuning for separate PR\r\n\r\nEventually doc changes etc but lets get the code working first.\r\n\r\nIf @VahidooX @okuchaiev can do code reviews again please do!! I'll update with mozilla results as soon as I can.", "Great review @VahidooX super thorough which is what I needed. Will take time to address everything but will get there!", "This pull request **introduces 1 alert** when merging 13f689cf01bf7881f661762cac4798bc16ef59bd into 4efd886121e5dd5fe25792116fd0c6afca036b98 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2fa6f899df9c5b3aba9f8c04d2888ac844cc6e06)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 1f941e7a14b0f5d1898927dfd37a4546f08ca5a4 into cd47137817e6d686f7551db18fd35610fd442343 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-1f480ecb9864b59c383b6e2bacf8608125b36b58)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 02b499bb39793d59712f57bdd9c2c745481fb6da into 1c24a5d35a9680b7436f505a8e69dc8f8373f3c2 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-dcd8e800d90d66d0fb9649e3f868cfc198f23642)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 4e97e126b423bcb888469b43397ed7c4dd0e9f19 into 51092ef7a9810ab096148cdf2d44ddfe6c72d028 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-3e11f6b66b14676dc90d9791f2eb2d861dbe1e71)\n\n**new alerts:**\n\n* 1 for Unused import", "Let's try get this over the finish line! @VahidooX I tried my best to answer as much as I can from the review but a lot became outdated. Would you be able to carry out another review on top?\r\n\r\nI also tried my best to clear out some of the hydra config stuff and expose them all in the yaml file but you may have a better idea on how to do so, please lemme know!\r\n\r\nFinally I'd like to get rid of the `wav2vec_base` class, which is only needed because there is no base class currently that only contains the data loading aspect of `ctc_models.py`. If we make `ctc_models.py` inherit from a similar class, then I think I'll be able to use that base class. Any suggestions?", "This pull request **introduces 1 alert** when merging 569a5c94c05314f21b98a441e25ef41c95db18ef into 4f46bd84d162da80a310c69f72980de935c1e99a - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-6522d0e1c0b32401060e1adfea8004a59f00c03e)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging 9f760fb9f347b158889b6e5ad080a7ae3e8f5017 into 4f46bd84d162da80a310c69f72980de935c1e99a - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-bb7528edcc01028ef93ae413a89a5d77d2532c9e)\n\n**new alerts:**\n\n* 1 for Unused import", "This pull request **introduces 1 alert** when merging cfeec2b466b2e4219777792c3063f53a8cb1026d into 4f46bd84d162da80a310c69f72980de935c1e99a - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-166680e05e4b47e397e78d39247381b264885d15)\n\n**new alerts:**\n\n* 1 for Unused import" ]

[]

[]

[ "Let's merge #1345 into your PR" ]

[]

[]

[]

[]

[]

[]

[ { "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" } ]

[]

[ "@Fackor I had this error while working with `pytorch_lightning==0.9.0`. Upgrading it to the `pytorch_lightning==1.0.3` solved the issue for me.\r\n\r\nIf you still want to work with `pytorch_lightning==0.9.0`, you should change `accelerator: ddp` to `distributed_backend: ddp` in [the config file](https://github.com/NVIDIA/NeMo/blob/main/examples/nlp/glue_benchmark/glue_benchmark_config.yaml)\r\n\r\n```\r\n # accelerator: ddp\r\n distributed_backend: ddp\r\n```\r\n\r\nHope that this helps :) ", "Hi, thanks for the issue. We have been working to pin the tutorial notebooks linked from the README to the correct version of NeMo and the correct version of the notebook. These changes are going into the next minor release. \r\n\r\nPlease try the notebooks here: https://github.com/NVIDIA/NeMo/tree/v1.0.0b2 and let us know if you still have issues.\r\n\r\n" ]

[ { "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 }, { "login": "VahidooX", "id": 23551647, "node_id": "MDQ6VXNlcjIzNTUxNjQ3", "avatar_url": "https://avatars.githubusercontent.com/u/23551647?v=4", "gravatar_id": "", "url": "https://api.github.com/users/VahidooX", "html_url": "https://github.com/VahidooX", "followers_url": "https://api.github.com/users/VahidooX/followers", "following_url": "https://api.github.com/users/VahidooX/following{/other_user}", "gists_url": "https://api.github.com/users/VahidooX/gists{/gist_id}", "starred_url": "https://api.github.com/users/VahidooX/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/VahidooX/subscriptions", "organizations_url": "https://api.github.com/users/VahidooX/orgs", "repos_url": "https://api.github.com/users/VahidooX/repos", "events_url": "https://api.github.com/users/VahidooX/events{/privacy}", "received_events_url": "https://api.github.com/users/VahidooX/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "Thanks for starting this! Yes, wave2vec 2.0 looks very intriguing and is something we wanted to try but did not have a chance yet.\r\n\r\nTo answer your questions:\r\n\r\n1. Yes! We will be happy to see your wave2vec 2.0 PR. Please have a look at https://github.com/NVIDIA/NeMo/blob/main/CONTRIBUTING.md and reach out to us if you have any questions. NeMo was never intended to only support NVIDIA-based research, so external models are very welcome too.\r\n\r\n2. I pretty sure we'll be able to help with code and pre-training some models, so that others can built upon them.", "Thanks @okuchaiev! Another question I had is the collection this belongs to. Technically the pre-training only requires unlabelled audio, and then the subsequent model can be fine-tuned for ASR, but I think all the code should live in the ASR collection for the time being. Was curious if there was any disagreement here", "ASR collection sounds like the right fit to me.", "Hi Sean! Please open a WIP PR with what you have. I would help you with any questions or problems.", "Will do :)", "Exciting! Looking forward to this and NVIDIA's pre-training + noise augmentation. ", "@SeanNaren will the pretrained weights shared by fairseq be compatible with your implementation on NeMo? @okuchaiev if not will NVidia provide one in NGC?\r\nP.S. It might be worth looking into https://arxiv.org/abs/2010.10504 (using conformer layers in a wav2vec2 training) as well.", "I just wanted to say that this will be really great for the community -- very much looking forward to seeing this in NeMo!!", "> @SeanNaren will the pretrained weights shared by fairseq be compatible with your implementation on NeMo? @okuchaiev if not will NVidia provide one in NGC?\r\n> P.S. It might be worth looking into https://arxiv.org/abs/2010.10504 (using conformer layers in a wav2vec2 training) as well.\r\n\r\nEventually I hope we create our own pre-trained models so we're not tied to the fairseq implementation. I feel like there are lots of cleanliness issues with the codebase, once I get ASR fine-tuning working, I plan to make a few clean-ups to the code for readability.", "I can't promise, but I think it is very likely we'll provide pre-trained checkpoints.\r\nAlso, @VahidooX is currently adding Conformer-like ASR model", "This might need to be re-opened, we're still missing the fine-tuning piece!", "The community greatly supports your integration of Wav2vec into Nemo.\r\nAny plan for releasing the update ?", "Hey! apologies on the radio silence, I haven't had a chance to do the fine-tuning port, but hoping to get some time this week to bring the code as a separate PR. This would complete the code integration however models would still need to be made.\r\n\r\nTies well with this dataset however https://ai.facebook.com/blog/a-new-open-data-set-for-multilingual-speech-research/ :) ", "That's excellent !\r\nAnd you link is an \"Ali Baba Cave\" !", "Hi any updates on this ?", "waiting for the implementation ..", "This will be added soon in #3297 but we are still training a few models to make sure the convergence is correct. " ]

[ { "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" } ]

[ { "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 } ]

[ "This works to me on the latest master:\r\n```\r\n model = nemo_asr.models.EncDecCTCModel.from_pretrained(model_name='QuartzNet15x5Base-En')\r\n```\r\nI think we had a fix for strict=False merged recently. Could you please try again (without strict=False) and re-open the bug if the issue still persists", "Just downloaded latest master again and tried to get the model still the same issue:-\r\n\r\nTraceback (most recent call last):\r\n File \"examples\\asr\\offline_asr.py\", line 9, in <module>\r\n asr_model = nemo_asr.models.EncDecCTCModel.from_pretrained(model_name='QuartzNet15x5Base-En')\r\n File \".\\nemo\\core\\classes\\common.py\", line 406, in from_pretrained\r\n strict=strict,\r\n File \".\\nemo\\core\\classes\\modelPT.py\", line 253, in restore_from\r\n instance = cls.from_config_dict(config=conf)\r\n File \".\\nemo\\core\\classes\\common.py\", line 258, in from_config_dict\r\n instance = cls(cfg=config)\r\n File \".\\nemo\\collections\\asr\\models\\ctc_models.py\", line 115, in __init__\r\n dist_sync_on_step=True,\r\nTypeError: Can't instantiate abstract class WER with abstract methods forward\r\n\r\nThis time I tried without 'strict'. Not sure what is going wrong here.\r\n", "Could you try `asr_model = nemo_asr.models.EncDecCTCModel.from_pretrained(model_name='QuartzNet15x5Base-En', strict=False)`", "@titu1994 I have tried it with strict=False and without strict parameter. Both are giving errors as mentioned in previous comments. Please check.", "@titu1994 @okuchaiev .. Tried it with both source code and pip installation, both Windows and Linux, both strict=True and False.. everywhere same error:-\r\n\r\n asr_model = nemo_asr.models.EncDecCTCModel.from_pretrained(model_name='QuartzNet15x5Base-En', strict=False)\r\n File \"C:\\Users\\e5602894\\Desktop\\NeMo-main\\env_files_nemo_main\\lib\\site-packages\\nemo_toolkit-1.0.0b2-py3.7.egg\\nemo\\core\\classes\\common.py\", line 406, in from_pretrained\r\n strict=strict,\r\n File \"C:\\Users\\e5602894\\Desktop\\NeMo-main\\env_files_nemo_main\\lib\\site-packages\\nemo_toolkit-1.0.0b2-py3.7.egg\\nemo\\core\\classes\\modelPT.py\", line 382, in restore_from\r\n return cls._default_restore_from(restore_path, override_config_path, map_location, strict)\r\n File \"C:\\Users\\e5602894\\Desktop\\NeMo-main\\env_files_nemo_main\\lib\\site-packages\\nemo_toolkit-1.0.0b2-py3.7.egg\\nemo\\core\\classes\\modelPT.py\", line 303, in _default_restore_from\r\n instance = cls.from_config_dict(config=conf)\r\n File \"C:\\Users\\e5602894\\Desktop\\NeMo-main\\env_files_nemo_main\\lib\\site-packages\\nemo_toolkit-1.0.0b2-py3.7.egg\\nemo\\core\\classes\\common.py\", line 258, in from_config_dict\r\n instance = cls(cfg=config)\r\n File \"C:\\Users\\e5602894\\Desktop\\NeMo-main\\env_files_nemo_main\\lib\\site-packages\\nemo_toolkit-1.0.0b2-py3.7.egg\\nemo\\collections\\asr\\models\\ctc_models.py\", line 139, in __init__\r\n log_prediction=self._cfg.get(\"log_prediction\", False),\r\nTypeError: Can't instantiate abstract class WER with abstract methods forward\r\n\r\nNot able to reopen the issue as closed by contributor. Please reopen and help in resolving it.", "The WER class does not have forward method anymore that we implement - https://github.com/NVIDIA/NeMo/blob/main/nemo/collections/asr/metrics/wer.py#L62-L181. \r\n\r\nSame for b2 branch. Please use reinstall.sh at the root of the Nemo directory and update pytorch lightning to 1.0.5 and try again.", "@titu1994 .. upgraded to pytorch-lightning==1.0.5 and the issue is gone. Thanks a lot for your help. 👍 Closing the issue." ]

[]

[]

[ "Hey @catskillsresearch, I don't have a solution to your problem but I rather have a question for you. \r\nI am also working on Unicode range characters, like yours. When I am training on a pretrained model, the reference string I am getting while training is empty. Did you face the same issue, how did you resolve it. If not, could you please tell how did you go about this transfer learning? \r\nthanks.", "Hi @gopesh97 I had a[ similar issue](https://github.com/NVIDIA/NeMo/issues/1283). NVidia folks gave some help. Here is what I do now to start a new language. \r\n\r\nTo use pretrained QuartzNet15x5, your training data needs to be 16000Hz. If that's not the case then start by resampling your data.\r\n\r\nThen, you need to copy and modify the YAML file `./NeMo/examples/asr/conf/config.yaml\r\n` to replace the English vocabulary with your new one. For example for Vietnamese the top of the file then becomes\r\n\r\n```\r\nname: &name \"QuartzNet15x5\"\r\nsample_rate: &sample_rate 16000\r\nrepeat: &repeat 1\r\ndropout: &dropout 0.0\r\nseparable: &separable true\r\nlabels: &labels [' ', \"'\", '-', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',\r\n 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',\r\n 'X', 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k',\r\n 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x',\r\n 'y', 'z', 'Á', 'Â', 'Ú', 'à', 'á', 'â', 'ã', 'è', 'é', 'ê', 'ì', 'í', 'ò',\r\n 'ó', 'ô', 'õ', 'ù', 'ú', 'ý', 'ă', 'Đ', 'đ', 'ĩ', 'ũ', 'ơ', 'ư', 'ạ',\r\n 'Ả', 'ả', 'ấ', 'ầ', 'ẩ', 'ẫ', 'ậ', 'ắ', 'ằ', 'ẳ', 'ẵ', 'ặ', 'ẹ', 'ẻ',\r\n 'ẽ', 'ế', 'ề', 'ể', 'ễ', 'ệ', 'ỉ', 'ị', 'ọ', 'ỏ', 'ố', 'ồ', 'ổ', 'ỗ',\r\n 'ộ', 'ớ', 'ờ', 'ở', 'ỡ', 'ợ', 'ụ', 'ủ', 'ứ', 'ừ', 'ử', 'ữ', 'ự', 'ỳ', 'ỷ', 'ỹ']\r\n```\r\n\r\nThen load the first time the pretrained model:\r\n\r\n```\r\nmodel = nemo_asr.models.EncDecCTCModel.from_pretrained(model_name=\"QuartzNet15x5Base-En\")\r\nyaml = YAML(typ='safe')\r\nyaml_path='new_yaml.yaml'\r\nwith open(yaml_path) as f:\r\n params = yaml.load(f)\r\nmodel.change_vocabulary(new_vocabulary=params['labels'])\r\n```\r\n\r\nSet up the checkpointing as follows:\r\n\r\n```\r\nimport pytorch_lightning as pl\r\nimport os, datetime\r\n\r\nclass ModelCheckpointAtEpochEnd(pl.callbacks.ModelCheckpoint):\r\n def on_epoch_end(self, trainer, pl_module):\r\n metrics = trainer.callback_metrics\r\n metrics['epoch'] = trainer.current_epoch\r\n trainer.checkpoint_callback.on_validation_end(trainer, pl_module)\r\n\r\npid=os.getpid()\r\ndt=datetime.datetime.now().strftime(\"%Y%m%d_%H%M%S\")\r\n\r\ncheckpoint_callback = ModelCheckpointAtEpochEnd(\r\n filepath='vietnamese_'+f'{dt}_{pid}'+'_{epoch:02d}',\r\n verbose=True,\r\n save_top_k=-1,\r\n save_weights_only=False,\r\n period=1)\r\n\r\ntrainer = pl.Trainer(gpus=[0], max_epochs=1000, amp_level='O1', precision=16, checkpoint_callback=checkpoint_callback)\r\n```\r\nSubsequently, when you load .CKPT files for retraining, that will look like this:\r\n\r\n```\r\nmodel = nemo_asr.models.EncDecCTCModel.load_from_checkpoint(last_fn)\r\n```\r\n", "@catskillsresearch why is it necessary to declare upper and lower case characters in label and use pretrained model which is better than train model from scratch? I'm also experimenting with Vietnamese.\r\nThanks!\r\n", "Not necessarily better. I am just pulling the letters out of the NIST transcription. If there are upper case letters it's because NIST threw them into the transcription somewhere. \r\n\r\nHere is the official position from NIST:\r\n\r\n> We will use caseless scoring for ASR. Sclite will convert both reference and system to a single case before alignment.\r\n\r\nKnowing that, next time around, I would take the uppercase characters out of the vocabulary and lowercase the ground truth, to reduce the size of the model. I don't expect it to have a big effect on results though.", "yeah. you can try converting phonemes to grapheme . have better results than charater level. how many hours of your data", "I don't have ground truth phonemic transcription. I have audio with grapheme transcription. The whole point of these fancy new neural net architectures is to cut out the middleman, i.e. go Audio -> Grapheme directly rather than Audio -> Phoneme -> Grapheme. That way we can just drop the study of phonemes altogether. Phonemes are old school.\r\n\r\nAs far as #hours, [NIST gave us 20](https://www.nist.gov/itl/iad/mig/openasr-challenge). 20 hours for 2020. That's enough, right? Maybe with a little data augmentation? And that transfer learning?", "@catskillsresearch Could you pls send me the link of Vietnamese ASR dataset?", "Nice!\r\nCan you give me the OpenASR2020 dataset? I just registered but the website is closed and can not be downloaded again. If possible, please send me the link to my email: [email protected] and maybe i will also share with you my dataset. I use it for research purposes only. \r\nThanks very much!", "@dangvansam98 and @hoangtuanvu, to get this dataset, first register for the [NIST OpenASR20 Challenge](https://sat.nist.gov/openasr20) by following [these instructions](https://sat.nist.gov/openasr20#tab_register). However, the registration deadline has expired. The evaluation period is about to begin. You can email NIST at [email protected] to ask them to let you in, they might say Yes. \r\n\r\nI can't give you the data myself because of the data usage agreement. You have to sign this agreement and submit it to them as a later part of the registration process.", "@catskillsresearch Thanks for your suggestion" ]

[]

[]

[ "Thanks Tom,\r\nat this moment it's unavoidable, but let's talk to Vlad - the owner", "What are the questions to me? ", "How can we instantiate the model without providing so much data?", "To make inference with a trained model basically we need only Dictionary of Intent and Slots, to know how to interpret the result numbers. They are 2 small text files stored in the data directory that is used for training. I can also add them inside of Config files, if it is better. For testing of training we need some training data as for any other model.", "No, just to build a graph and export it to ONNX. " ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ "This pull request **introduces 1 alert** when merging eaf6cf589e86f83acb2f48c1e6dc831e1533af89 into cd55594e0f60a4bcbaf6d0e4196c1404591e0076 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-9ee197eb87b278c5e13fc11f091b00114da5f04b)\n\n**new alerts:**\n\n* 1 for Unused import" ]

[]

[]

[]

[]

[]

[ "This pull request **introduces 1 alert** when merging 77f3867cb8ed29de19137c14b7606acbd6ed58fd into 53b31f7b902f96e3b964214373c328b3fa2d3606 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-fdbaf1bef5bfb2c8b4b97e7eda74c8c0cb5d0a77)\n\n**new alerts:**\n\n* 1 for Unused import" ]

[]

[]

[]

[]

[]

[ "This pull request **introduces 1 alert** when merging 356ec3a4b14e2c9877fc393d78b1887de38e7056 into 730da9886249c8f4acc766b0bde516a14681ac00 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2e81cf2c4aafdecb672b604ea76fcae4bf8848ba)\n\n**new alerts:**\n\n* 1 for Variable defined multiple times" ]

[ { "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 am facing similar issues using unicode, can you share some example lines from your manifest? One possibility is that you need to use `ensure_ascii=False` as an option to json.dump when generating the manifests: `json.dump(metadata, f, ensure_ascii=False)`", "@gopesh97 have you got any success on Hindi. I'm working on urdu , just curious to know whether this Nemo model works for right to left language?", "Ensure that you have the correct language settings in your environment (native or in container). Most ascii encoding issues can be solved by setting the `LANG` and `LC_ALL` env variables to a UTF-8 default" ]

[]

[]

[ "I think you need to wait for this PR to merge first https://github.com/NVIDIA/NeMo/pull/1319\r\n\r\nBased on what I saw in this notebook https://github.com/NVIDIA/NeMo/blob/73875d874b006b672a8e407687ef414ef7a7b282/tutorials/asr/Offline_ASR.ipynb" ]

[]

[]

[ "This pull request **introduces 1 alert** when merging 02d1de2a5df5480faf148a0623fcfa60089a3cff into c09012b6018243861a59a51401393bc8d64c163b - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-7e2ade19bb3ce1e9ccec72842028c33c8fea56c0)\n\n**new alerts:**\n\n* 1 for Unused import", "@blisc `_target_` is required as a validation. The config merge explicitly fails when trying to merge with the yaml override config otherwise. \r\n\r\nAs to why I wrote the class explicitly in string, that can be corrected by import + helper function. I will do that in next iteration. ", "This pull request **introduces 2 alerts** when merging 31a354d69ec750100ef8e892dbce87af36e354df into cd55594e0f60a4bcbaf6d0e4196c1404591e0076 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-a407e6697580ef067f80b8541e67e118f40c77dc)\n\n**new alerts:**\n\n* 2 for Unused import", "@ericharper @tkornuta-nvidia @blisc Please find two example scripts with proposed hybrid and full config structure. Also updated readme.", "This pull request **introduces 2 alerts** when merging f7017c4ede0b4e7c2a8499ccdbb56b2f82680398 into d28114ae203558a0bd1bcc69e78aeb957ea9ed5e - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-d44408e5b20dfc1a965eaa3a143c853dea910a30)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging add0e9a8b287127294ded478859f1e2d51989304 into d28114ae203558a0bd1bcc69e78aeb957ea9ed5e - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-9dae3b1b4ac42e4d1793ebbd34c21293bae4aca8)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 4976f68907f5675ec24dfdf2d009d760c2507dfb into d28114ae203558a0bd1bcc69e78aeb957ea9ed5e - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-3c969fd465ba1b2fad4c01d869c747362d03ee48)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 43b8e7632aae7d162231458b436028bbc452bf96 into e82ff1863bf750673400592a8220d63dff51933a - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-c199742b48b43d34c2b74c2b40779000d3e38318)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 2d91036768bdd1341d89b831744798e9b1354358 into cb0002f29c847915e5104aa780d5ebe49ec16f95 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-202df967a093917201b02ebedc54cfb2ec80e013)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging f38f36e938a7f6cecae39f23d73c84886e726b88 into fb802c524d784da616e7fe7788255ef89520682d - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-92796a87651732adfe270e04f672dc2af4013012)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 8757286b9f34ae9d555592bf296a3d7ead217ca2 into 30b2ed2a84b5e568d15b97079b15a9559450f480 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2b138e9caefb3613dfe2a292aac3be2cf1576bd9)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 3 alerts** when merging ea66b30e34b0333158295383cb277dc368fae5d1 into 30b2ed2a84b5e568d15b97079b15a9559450f480 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-2f0cfb0066e0600f9a9233587b9cbaa393038ac7)\n\n**new alerts:**\n\n* 3 for Unused import", "This pull request **introduces 2 alerts** when merging 3aeba48af08a79bb806644590cf0872f2fb053ac into 30b2ed2a84b5e568d15b97079b15a9559450f480 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-1188f81522a773470787ad486303dd319e9d51f2)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 83296229e02eb8385c2277a6d85423763b185179 into 30b2ed2a84b5e568d15b97079b15a9559450f480 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-17cb2f0cc2c96c8ed8db821489d61750de596b6e)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 06e6cdd7ec28d3ddc881344bdc8798d2f6485a1c into 4e804fd88bc197d031e97972957d8ad21e0c002f - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-c067612c28552566e7cf5d388ceca2a288b62620)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging f7c592d251de2612974b049e3f5b219a70495225 into 5572c82c0ba0c160eb940e94c2954886203abba6 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-74021030c1049dfcbdccb0e5ec0d75411589519a)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 73e3f05783d269f357350969469f5a9d7d33b395 into 17357c0752b6c323644b17ca62188830ca53cc85 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-5ef467d5822933235cdb162c26a78950404d984f)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 869ba22adbe71d0415e0d07fc6c4bae75dd7ba4d into b38af60fec6fe0bc527eb88413306ff7713251d9 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-215bb908060ccee2e180bdd7bfeae3ddfc232e2c)\n\n**new alerts:**\n\n* 2 for Unused import", "This pull request **introduces 2 alerts** when merging 56ecbff74790ff5daa83ea23e225bcfa38a8d428 into 6ae02c91eadcf0293d767597231ea7803aca8ce6 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-36b94bb6e80340d9580be92b88ff5fd8abe3e096)\n\n**new alerts:**\n\n* 2 for Unused import" ]

[]

[]

[ "This pull request **introduces 3 alerts** when merging 428a72d3463ecaa1d11dee86742ade414cdf67b5 into 4e4597a85d80c77603102603e35fb033ba852c64 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-7a79f4903246158dc860d19779ae73ee01878006)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method\n* 1 for Unused local variable", "This pull request **introduces 3 alerts** when merging a4c4c79b4c7cb33c2da18a37810395c263567a98 into 4e4597a85d80c77603102603e35fb033ba852c64 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-6eb166b4e7a827d49375e814b9a7303a8ca8fd3c)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method\n* 1 for Unused import", "This pull request **introduces 3 alerts** when merging 2c900aa2210681fd54480024469dcdc0af1ea802 into 4e4597a85d80c77603102603e35fb033ba852c64 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-7a71258607c4232a4dca02957aef00ac7c6848d5)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method\n* 1 for Unused import", "This pull request **introduces 2 alerts** when merging ac86cb3b4c4de575e84503a732df7492248d3fd4 into c09012b6018243861a59a51401393bc8d64c163b - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-0d28390d187cb353dcce38b4ec171c4bf6219aa4)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 75aa1e4754ad21e689693ca56a1c4d12194f595d into c09012b6018243861a59a51401393bc8d64c163b - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-f5b4aad8feb11218c73263d10a631fa21d46cc80)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 25fff7f7a5b080b21cba7d6f07353435c803e047 into c09012b6018243861a59a51401393bc8d64c163b - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-ce0f2eec0f001876b6acc2b673d4740c0465fe81)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 09d217c72f92ae9a2df77a454834f69fff413646 into c09012b6018243861a59a51401393bc8d64c163b - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-a91b4190f79dc5e0805a28aff41cbb1f2f4742ec)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging cda38ae4df75c0be717278d93ff0cd60013afd28 into c09012b6018243861a59a51401393bc8d64c163b - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-ce08f86a99e226ac0eb7bb432ff21271f9d41987)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging c24b0f697722bf781b0c6680d6f096d12acc7ff7 into 53b31f7b902f96e3b964214373c328b3fa2d3606 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-e78bdd4be32e9c4742c448b9d70d6844c8d2f65c)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 7a619052ec03643f2ae77fdb74749bcbe76458b8 into 53b31f7b902f96e3b964214373c328b3fa2d3606 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-af06684a77f4f9f6bccc26f36a62022bd919184d)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging ef7a23672feda0be9bd1bb3402f4664e40627131 into 53b31f7b902f96e3b964214373c328b3fa2d3606 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-f5b8bd1a2758b8b210f444db3a9432070c68168a)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 9c4993dd1bf0d3e335cb2a5a5079ee0f1814f009 into cd55594e0f60a4bcbaf6d0e4196c1404591e0076 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-3a5101717384f083ad5cb202b532464cf6da442f)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging fb324d033bec62286263665f4e89387259d7a3c1 into cd55594e0f60a4bcbaf6d0e4196c1404591e0076 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-70d0e489934e832ccb58b5d29e047f309640177e)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 2473812f8c73d3d5b4136430e9807214796de87c into cd55594e0f60a4bcbaf6d0e4196c1404591e0076 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-34c5eaa7d5480b343eaa854ee4aaae59dac289db)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 165359935df936c5aee631e5ff0044ddddb8e56b into cd55594e0f60a4bcbaf6d0e4196c1404591e0076 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-63466207f537576283a1dad48cd95606d94c4e37)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging 81dff89ee556df1a2367e3b19b792b4e46200c4f into 7f0d33bd44a22214970162986d490706ea63e1b2 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-7c2877114a7d6e552553fd233c8fb073814be8a5)\n\n**new alerts:**\n\n* 2 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging ea93557ad4e22a503a99c6f6cb66ffd2049041a5 into 89579d87649c894fd5c74892308d4ac0b07dec4d - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-4eebf35187172454febb826cb35192d24198464c)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging f6787c0a302feb6998490f3fddf151cc3505662a into 89579d87649c894fd5c74892308d4ac0b07dec4d - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-aacb5c910e5f9f7d9d839eb1769e01701d64d7e6)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging 7c0fe3684bda17527e5a005bc6f2a29216ea9aca into d061dfd7970b95790d713b24f7874511129833e5 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-77680444813b2e749f4b73c18929aa5dc332cde0)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging a7c72073b342dfe35a18cadde0b7ce373d1c541b into d061dfd7970b95790d713b24f7874511129833e5 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-c0e9ef67586bc52e23de77dcf5e284554783e0ea)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging acfb8e6721378d4310de1a89e9fc03acb935829a into d061dfd7970b95790d713b24f7874511129833e5 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-217925a6759d33c93302d210c0aa0d2871985c75)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging fd8764a5d51723e1ca7da895437b8e9efbbf0507 into d061dfd7970b95790d713b24f7874511129833e5 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-7e6df0a5f9cebf2a768c9ae29a294bcc2987b2e3)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging 5f241d467acc436b36b538d6fd72f178d6ed402b into 2da227a757857fcbd578bad3ba6292de2230ea3d - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-172cfdf5dff8f7bfec37071a725ffd14a4f81559)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging f0cae87e9702662c2fd4b9c3f091f113be482544 into 48620bb1369dc72dcac7b9a4e75b113f1cc55d36 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-de58d42dafe15311c1126fa3c329ab833bed59ef)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging b625c95939c286e7dd8afab4bed8257caf07f083 into f444989d44d1201aa12be3269d2262f1908bf8b8 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-b2bb71005c0bd4a6fcf5e2472e12130de6183caf)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging fe4083938d33a483394139bc7c1abcf52dd57965 into 17fc2541172dd3d781bfa0b5a1fa41576d4bfe21 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-f4e7d081f7e837b89165173d3b97e2461a0ddbbb)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging c6b5b3c567d613f6be8b93fc9f9989387a4b993c into 17fc2541172dd3d781bfa0b5a1fa41576d4bfe21 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-dc988753ad20ee589f9cf8d48afbcb385cde4bfd)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging 8ea849b0bd2558c3da12de27e1f2c347564c4e75 into 17fc2541172dd3d781bfa0b5a1fa41576d4bfe21 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-4fe8ec272743c7766bab4120b83bf3dcb0365824)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 1 alert** when merging be0887825351eaff026ab0d4820cb1f7e263a5c2 into 17fc2541172dd3d781bfa0b5a1fa41576d4bfe21 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-707a119a1735b6c6488a3b8a344d8627ad28a5f3)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method", "This pull request **introduces 2 alerts** when merging d1ed6402c74b1061b253ceb4e4678a9e3be5ddf9 into 17fc2541172dd3d781bfa0b5a1fa41576d4bfe21 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-ae2454380e1a614fd45388c495e61b243588146b)\n\n**new alerts:**\n\n* 1 for Signature mismatch in overriding method\n* 1 for Unused import" ]

[]

[]

[]

[]

[]

[ "This pull request **introduces 3 alerts** when merging 916b8c79da313e93bf92ab7fed0d2f0e5436b2d9 into 730da9886249c8f4acc766b0bde516a14681ac00 - [view on LGTM.com](https://lgtm.com/projects/g/NVIDIA/NeMo/rev/pr-53d748f2d12e33432e771ec04560e9d2cba50fb0)\n\n**new alerts:**\n\n* 2 for Unused import\n* 1 for Unused local variable" ]

[]

[]

[ "@okuchaiev that won't work sadly with multiple data loaders. Each will be tagged with a specific name prepended so exp manager cannot infer it. At least not unless we also attach a \"default\" validation wer to our logs \r\n\r\nWe actually do attach a default \"val_loss\", with the default as the first dataloader passed to the validation filepath (can be changed with a flag). So we can add some logic to our wer logging to add a default wer to the logs too. ", "So can this: \"modify ASR configs (for speech recognition models only) to monitor: \"validation_wer\" instead of \"val_loss\" by default\" be done for a case of single data val loader at least (which is what our configs are)?\r\n" ]

[]

[]

[]

[]