Datasets:

---

renwei2024

/

nemo-github-issues

like 0

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null } ]

[]

[ "If you want to finetune using the original tokenizer, yes you'll need to normalize all numbers to spoken words. \r\n\r\nChanging tokenizer means you'll need a large amount of data to retrain the model, that is not suggested unless you have several thousand hours of speech to reach best results ", "> If you want to finetune using the original tokenizer, yes you'll need to normalize all numbers to spoken words.\r\n> \r\n> Changing tokenizer means you'll need a large amount of data to retrain the model, that is not suggested unless you have several thousand hours of speech to reach best results\r\n\r\nHow to use the original tokenizer? \r\nI also created [a discussion](https://github.com/NVIDIA/NeMo/discussions/9217) for this.\r\n", "This issue is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.", "This issue was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" } ]

[]

[]

[]

[]

[]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[ { "login": "gdengk", "id": 160076886, "node_id": "U_kgDOCYqUVg", "avatar_url": "https://avatars.githubusercontent.com/u/160076886?v=4", "gravatar_id": "", "url": "https://api.github.com/users/gdengk", "html_url": "https://github.com/gdengk", "followers_url": "https://api.github.com/users/gdengk/followers", "following_url": "https://api.github.com/users/gdengk/following{/other_user}", "gists_url": "https://api.github.com/users/gdengk/gists{/gist_id}", "starred_url": "https://api.github.com/users/gdengk/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/gdengk/subscriptions", "organizations_url": "https://api.github.com/users/gdengk/orgs", "repos_url": "https://api.github.com/users/gdengk/repos", "events_url": "https://api.github.com/users/gdengk/events{/privacy}", "received_events_url": "https://api.github.com/users/gdengk/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "Hi @ericharper can I get a review on this PR? Thanks!" ]

[]

[]

[]

[ { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" } ]

[]

[]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "/cicd", "/cicd", "/cicd" ]

[ { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "CICD passed, only optional test did not pass. Can go ahead & merge" ]

[ { "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" } ]

[]

[ "[all] was never guaranteed to work on Windows. \r\n[asr] can and should work on windows just fine. \r\nYttm has long been removed from NeMo #8322 " ]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null }, { "id": 4811884691, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ikw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/ASR", "name": "ASR", "color": "F3B33E", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "I will modify the code so that all the test cases pass soon. ", "Thank you for suggesting this idea.\r\nCurrently I don't have bandwidth to test this so let me check on this as soon as I have bandwidth to take care of this. \r\nThis needs to be tested with long-form and short-form audio clustering. ", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale.", "@tango4j @nithinraok Reminder to please review it ", "Hi. @AatikaNazneen Can you provide 4 hour long audio data for testing that is creating the OOM?\r\nAlso, there is a limit in the audio length size. Please be aware that the NeMo speaker diarization cannot handle infinite length audio. Even if it is long form clustering method, the length of audio is capped by GPU memory. ", "haven't tested this file but an example for 4 hours podcast that can used for testing..\r\nhttps://www.youtube.com/watch?v=HoRMD9NtKYc", "I used this one: https://www.youtube.com/watch?v=0OhTQnDp1yw\r\nI was able to do this diarize this on sagemaker g4 instance which has 16 GB GPU RAM and 16 GB CPU RAM after this change.", "@AatikaNazneen \r\nI was not able to find a way to download the wav file from your youtube link. Could you please extract wav file and upload it through Googledrive or dropbox account?\r\nAlso, reviewing this PR would take some time, since we need to test with other scenarios. \r\nPlease provide us with the long form audio files you can share, then it will expedite the process to test your PR. \r\n", "@tango4j You can use [this ](https://traffic.libsyn.com/secure/schoolofmotion/School_of_Motion_EOY_Podcast-AUDIO-20221228.mp3?dest-id=485169) audio or any audio. The main purpose is that the long form clustering was taking up more RAM previously and that has been optimized so it can be tested of any long audio but you would have to limit your resources to test the before and after effect.", "@tango4j Any updates?", " @tango4j @nithinraok ", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "@tango4j @nithinraok Reminder to review it please", "@AatikaNazneen \r\nSorry for the frustration. Let me review this today.", "Hi, @AatikaNazneen.\r\n\r\nI have run the original main version and your for-loop version.\r\nFirst of all, your analysis that `torch.tile` can be taking excessive memory is valid.\r\nHowever, we can let the `get_argmin_mat` by using `torch.expand`\r\n\r\n```python\r\nIn [1]: import torch\r\n\r\nIn [2]: vec = torch.rand(1000, 1).T.to(\"cuda:0\") # 556MB \r\n\r\nIn [3]: vec_tile = vec.tile(10000000,1) # 38704MB, 38148 MB was added \r\n\r\nIn [4]: vec_expand = vec.expand(10000000, 1000) # 0MB was added\r\n``` \r\nNow, if we change the `get_argmin_mat` line to these:\r\n```python\r\ncurr_mat = curr_scale_anchor.expand(base_scale_anchor.shape[0], -1)\r\nbase_mat = base_scale_anchor.expand(curr_scale_anchor.shape[0], -1).t()\r\n ```\r\n These are the time taken for three 10 min audio files.\r\n \r\n> torch.expand: 0.0266 seconds\r\ntorch.tile: 0.0568 seconds\r\nFor Loop: 0.1970 seconds\r\n\r\nBottom line is, we need to modify this to use expand, instead of adding for loop.\r\nFor loop dramatically increases the Time complexity from O(S) to O(NS) where S is number of scales and N is the frame length in base-scale.\r\n\r\nLet me push the `torch.expand` version of code and please check if there is still memory issue.\r\nAs you can see in the torch example I showed above, there should be no memory choke point with `torch.expand`.\r\n\r\n ", "@AatikaNazneen @tango4j are more checks waiting to be reviewed?", "@tango4j Your modification did not do the work for me and i still get OOM error.\r\nThe reason is that torch.expand doesnt increase memory but when we do the subtraction operation, it occupies the double the memory for both the vectors. The for loop reduces the memory consumption by breaking the problem into a smaller chunk so the peak memory occupied is much lesser as compared to without the for loop\r\n<img width=\"1231\" alt=\"Screenshot 2024-08-26 at 4 25 38 PM\" src=\"https://github.com/user-attachments/assets/3db9b738-f9de-4e6f-8e12-0d203e281a35\">\r\n\r\n", "@tango4j If you're convinced then please revert it back to the for loop version. and the rest of the labels needed to run all the checks as well.", "The time complexity is theoretically the same but using a for loop, the previous operation memory is freed (hence the increase in time) but it significantly reduces the space complexity. \r\nUsing the expand function, we need 2 (n^2) sizes at a particular memory and using the for loop we use n * (2n) memory but at a particular time, we only need 2n memory so we dont get OOM.\r\nLet me know if you need further clarification", "@AatikaNazneen \r\nI still do not get the idea of for loop and expand function getting the same complexity.\r\nBoth in theory and experiments, for loop is increasing the running time.\r\nIt is array assignments versus sequential process so for loop is obviously takes more time.\r\nand then if you implement it, it also shows slower speed.\r\n\r\nI missed the part where expand-approach causes memory consumption when it performs subtractions.\r\n\r\nSo in short, \r\nI think we should create an option to use forloop to handle extremely long audio.\r\nWe cannot increase the time complexity to xN steps just for rarely handled extremely long audio files.\r\n", "@tango4j We can go for that implementation when we use long-form-clustering pipeline.\r\n> I missed the part where expand-approach causes memory consumption when it performs subtractions.\r\n\r\nIf you take a look at line 13 in the image I shared, the memory increment is twice the amount increment in line 9\r\nAlso, I tested the 5 hour audio of sagemaker with mlg4dn and it was failing with the expand solution\r\n\r\n", "@AatikaNazneen \r\nI can see that line 13 in your screenshot shows the additional memory usage during the subtraction operation of the copied tensors. I neglected this part when I was testing.\r\nThis is why we are not saving much on memory peak usage compared to \"tile\" method.\r\n\r\nSo at the end of the day, I believe this is all about memory-time trade off.\r\nI think, as you suggested, we can let the system use \"for loop\" for long form audio clustering cases. That way we can only value memory efficiency for long form audio clustering and reduce the peak memory usage. \r\n\r\nI will add this option shortly to long form audio clustering path.\r\n- standard: expand (speed oriented)\r\n- long form audio: for loop (memory oriented)\r\n", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "When do you think all of this will be completed? \r\nThanks..\r\n@tango4j @nithinraok ", "@AatikaNazneen @tttalshaul \r\nSorry that I have been traveling overseas for conference. \r\nWill work on this PR starting today and finalize this week.", "reminding 🙏🏿", "@AatikaNazneen @tttalshaul \r\nI pushed the change to the @AatikaNazneen 's Repo.\r\nThis version is using expand for less than 10K embedding frames then use for-loop for more than 10K embeddings.", "@tango4j Thank you.\r\nWould it be an official option in main NeMo repo/pip package? ", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days." ]

[ { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "/cicd", "/cicd", "/cicd", "/cicd", "/cicd" ]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4847373924, "node_id": "LA_kwDOC_bI7s8AAAABIO0OZA", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/common", "name": "common", "color": "c5def5", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[]

[]

[]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null }, { "id": 4811884691, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ikw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/ASR", "name": "ASR", "color": "F3B33E", "default": false, "description": "" } ]

[]

[ "@titu1994 if you could rereview at your convenience, that would be appreciated.", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale.", "@galv I reopened this PR. It's not yet ready for review (I will let you know once I finalize it), but the code is now working, and the performance matches the \"greedy\" approach.", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale.", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null }, { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "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" }, { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null } ]

[]

[ "I've tried trained models from previous version of NeMo and newly trained within version 1.23.0 and etc.", "As I understand, problem is in: [nemo2riva/nemo2riva/patches/tts/fastpitch.py at r2.15.0 · nvidia-riva/nemo2riva](https://github.com/nvidia-riva/nemo2riva/blob/r2.15.0/nemo2riva/patches/tts/fastpitch.py). Runtime didn't catch this additional params and runtime-check failing", "hey?", "This issue is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.", "This issue was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null } ]

[ { "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 } ]

[ "Yes it's supported, please find installation instructions here - https://docs.nvidia.com/nemo-framework/user-guide/latest/nemotoolkit/asr/asr_language_modeling.html#n-gram-language-modeling\r\n\r\nThat bash script has to be run by users as is, rather than python to install flashlight ", "I followed this guide but it refers to scrips that don't exist anymore. For example the following script is not relevant already:\r\n```\r\npython eval_beamsearch_ngram.py nemo_model_file=<path to the .nemo file of the model> \\\r\n input_manifest=<path to the evaluation JSON manifest file \\\r\n kenlm_model_file=<path to the binary KenLM model> \\\r\n beam_width=[<list of the beam widths, separated with commas>] \\\r\n beam_alpha=[<list of the beam alphas, separated with commas>] \\\r\n beam_beta=[<list of the beam betas, separated with commas>] \\\r\n preds_output_folder=<optional folder to store the predictions> \\\r\n probs_cache_file=null \\\r\n decoding_mode=beamsearch_ngram \\\r\n decoding_strategy=\"<Beam library such as beam, pyctcdecode or flashlight>\"\r\n```\r\nbecause there is no eval_breamsearch_ngram.py [here](https://github.com/NVIDIA/NeMo/blob/stable/scripts/asr_language_modeling/ngram_lm/eval_beamsearch_ngram.py).", "@karpnv could you update docs ?", "NeMo of previous versions has it but there are no such parameters as `decoding_strategy`. I mean it can't be used as flashlight. And even in older versions of NeMo it's absent. For example in [1.18.0](https://github.com/NVIDIA/NeMo/blob/v1.18.0/scripts/asr_language_modeling/ngram_lm/eval_beamsearch_ngram.py)", "This issue is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.", "This issue was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "Tests passed: https://github.com/NVIDIA/NeMo/actions/runs/8977317099\r\nLet me know if you want to merge" ]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4811884691, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ikw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/ASR", "name": "ASR", "color": "F3B33E", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "To verify that latency does indeed fall, I ran the above prefixed by `nsys profile --env-var=NSYS_NVTX_PROFILER_REGISTER_ONLY=0 -t nvtx -c nvtx -p GALVEZ_START` with batch_inference=false and batch_inference=true. I added an nvtx range just for the decoder.\r\n\r\n\r\n\r\nbatch_inference=true case:\r\n```\r\nTime\tTotal Time\tInstances\tAvg\tMed\tMin\tMax\tStdDev\tStyle\tRange\r\n93.9%\t71.801 s\t1\t71.801 s\t71.801 s\t71.801 s\t71.801 s\t0 ns\tPushPop\tGALVEZ_START\r\n6.1%\t4.683 s\t460\t10.180 ms\t1.173 ms\t796.420 μs\t117.883 ms\t19.312 ms\tPushPop\tdecoder\r\n```\r\n\r\nbatched_inference=false case:\r\n```\r\nTime\tTotal Time\tInstances\tAvg\tMed\tMin\tMax\tStdDev\tStyle\tRange\r\n86.7%\t79.410 s\t1\t79.410 s\t79.410 s\t79.410 s\t79.410 s\t0 ns\tPushPop\tGALVEZ_START\r\n13.3%\t12.217 s\t460\t26.558 ms\t3.800 ms\t2.125 ms\t171.248 ms\t38.294 ms\tPushPop\tdecoder\r\n```\r\n\r\nYou can see that the decoder takes half the time it used to, and that its maximum runtime is 118ms, down from 171 ms.", "Will need to get back to you about what it takes to make this the default algo. But the main concern is that many existing \".nemo\" files will have the \"strategy\" field of the relevant config set to \"greedy\". It's not really clear to me how we can convert those users using pretrained models to use the \"greedy_batched\" strategy. For what it's worth, my tests check that the interfaces are effectively the same, so I could just use GreedyBatchedCTCInfer whenever \"greedy\" strategy is specified, unless that distresses you.", "I see CICD pipeline passed: https://github.com/NVIDIA/NeMo/actions/runs/8976148083\r\nLet me know if you need/ready to merge", "Hmm I see your point. Lets do this - \r\n1) Set default in config to greedy_batch\r\n2) Add logging message to greedy (onetime log) that informs users that greedy_batch strategy is faster. \r\n3) All newer models will automatically be using the greedy batch path then \r\n\r\nI'm not average to replacing \"greedy\" with greedy_batch, but then we need a different flag name to allow users to fall back to older codepath. Maybe we can call it \"greedy_legacy\" then?", "@titu1994 done. To clarify, I went with the warn-once path rather than making \"greedy\" pointed to the new implementation." ]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null }, { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null } ]

[]

[ "> Please capture the error and include in the ImportError so users will know exactly what's not being imported properly.\r\n\r\nImportError is raised at the model level not layer level. Most of the cases here are using Apex's `groupnorm`, I added the ImportError when instantiate the model, not for each layer we invoke groupnorm.\r\n\r\n@ericharper do you think this is sufficient ?", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 1923495043, "node_id": "MDU6TGFiZWwxOTIzNDk1MDQz", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/core", "name": "core", "color": "f25e3a", "default": false, "description": "Changes to NeMo Core" }, { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4418253970, "node_id": "LA_kwDOC_bI7s8AAAABB1k0kg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/TTS", "name": "TTS", "color": "49AF28", "default": false, "description": "" }, { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 1923495043, "node_id": "MDU6TGFiZWwxOTIzNDk1MDQz", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/core", "name": "core", "color": "f25e3a", "default": false, "description": "Changes to NeMo Core" }, { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null }, { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" } ]

[]

[ "cicd", "cicd", "cicd", "cicd", "/deploy", "cicd", "/deploy", "/deploy", "/deploy", "cicd", "/deploy", "/deploy", "/deploy", "/deploy", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "jenkins", "Pipeline passed https://github.com/NVIDIA/NeMo/actions/runs/8928355809" ]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null }, { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 1923495043, "node_id": "MDU6TGFiZWwxOTIzNDk1MDQz", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/core", "name": "core", "color": "f25e3a", "default": false, "description": "Changes to NeMo Core" }, { "id": 4556669155, "node_id": "LA_kwDOC_bI7s8AAAABD5lA4w", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/cherry-pick", "name": "cherry-pick", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4811884691, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ikw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/ASR", "name": "ASR", "color": "F3B33E", "default": false, "description": "" }, { "id": 4847373924, "node_id": "LA_kwDOC_bI7s8AAAABIO0OZA", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/common", "name": "common", "color": "c5def5", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "It causes a segmentation fault in GPU tests (but works fine in casual settings -- probably too many daemonic threads spawned in tests when enabled by default, or otherwise some global state conflict). Maybe I'll revisit this later but don't want to invest more time into it right now: closing." ]

[ { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "Merging since only test failure is 1 known problematic test." ]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "Pipeline passed: https://github.com/NVIDIA/NeMo/actions/runs/9018587937" ]

[ { "id": 1923495043, "node_id": "MDU6TGFiZWwxOTIzNDk1MDQz", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/core", "name": "core", "color": "f25e3a", "default": false, "description": "Changes to NeMo Core" } ]

[]

[ "@ShriyaPalsamudram Tested on Eos single-node, the updated code spits out a RuntimeError when the lock is not obtained within the given timeout period.", "How were you able to create a timeout situation in one node? Did you set the timer to very very small value?" ]

[ { "id": 1923495043, "node_id": "MDU6TGFiZWwxOTIzNDk1MDQz", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/core", "name": "core", "color": "f25e3a", "default": false, "description": "Changes to NeMo Core" }, { "id": 4811884691, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ikw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/ASR", "name": "ASR", "color": "F3B33E", "default": false, "description": "" } ]

[]

[ "One clean solution could be to add a method such as `.downcast(self, dtype=...)` to most modules that casts all parameters and buffers to a correct precision for a given dtype. It'd have to be recursively called so that every \"layer\" or \"block\" can define how it needs to be cast, e.g. `rnnt_model.downcast` calls `{encoder,decoder,joiner}.downcast`, which call `transformer_block.downcast`, etc etc.\r\n\r\nThis is under the assumption that simply calling `model.bfloat16` may cast some weights into a too-low precision. ", "#9198 supersedes this." ]

[ { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "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" }, { "id": 1923495043, "node_id": "MDU6TGFiZWwxOTIzNDk1MDQz", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/core", "name": "core", "color": "f25e3a", "default": false, "description": "Changes to NeMo Core" }, { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null } ]

[]

[ "Could we replace `self._lock` itself with the timeout-enabled one? The parent Apex distributed optimizer class also uses `self._lock` (e.g., [here](https://github.com/NVIDIA/apex/blob/master/apex/contrib/optimizers/distributed_fused_adam.py#L832)) and we want to catch those as well if they take too long.", "Testing a modified version in this draft PR: https://github.com/NVIDIA/NeMo/pull/9087", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4556669155, "node_id": "LA_kwDOC_bI7s8AAAABD5lA4w", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/cherry-pick", "name": "cherry-pick", "color": "ededed", "default": false, "description": null }, { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days." ]

[ { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null } ]

[ { "login": "JRD971000", "id": 71892896, "node_id": "MDQ6VXNlcjcxODkyODk2", "avatar_url": "https://avatars.githubusercontent.com/u/71892896?v=4", "gravatar_id": "", "url": "https://api.github.com/users/JRD971000", "html_url": "https://github.com/JRD971000", "followers_url": "https://api.github.com/users/JRD971000/followers", "following_url": "https://api.github.com/users/JRD971000/following{/other_user}", "gists_url": "https://api.github.com/users/JRD971000/gists{/gist_id}", "starred_url": "https://api.github.com/users/JRD971000/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/JRD971000/subscriptions", "organizations_url": "https://api.github.com/users/JRD971000/orgs", "repos_url": "https://api.github.com/users/JRD971000/repos", "events_url": "https://api.github.com/users/JRD971000/events{/privacy}", "received_events_url": "https://api.github.com/users/JRD971000/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" } ]

[]

[]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days." ]

[ { "id": 4556669155, "node_id": "LA_kwDOC_bI7s8AAAABD5lA4w", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/cherry-pick", "name": "cherry-pick", "color": "ededed", "default": false, "description": null }, { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "CI passed https://github.com/NVIDIA/NeMo/actions/runs/9387773801" ]

[ { "id": 1923495043, "node_id": "MDU6TGFiZWwxOTIzNDk1MDQz", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/core", "name": "core", "color": "f25e3a", "default": false, "description": "Changes to NeMo Core" }, { "id": 4418253970, "node_id": "LA_kwDOC_bI7s8AAAABB1k0kg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/TTS", "name": "TTS", "color": "49AF28", "default": false, "description": "" }, { "id": 4811884691, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ikw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/ASR", "name": "ASR", "color": "F3B33E", "default": false, "description": "" }, { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 4811888028, "node_id": "LA_kwDOC_bI7s8AAAABHs-VnA", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Speaker%20Tasks", "name": "Speaker Tasks", "color": "871AD4", "default": false, "description": "" }, { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 4847373924, "node_id": "LA_kwDOC_bI7s8AAAABIO0OZA", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/common", "name": "common", "color": "c5def5", "default": false, "description": "" }, { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null } ]

[]

[]

[ { "id": 4847373924, "node_id": "LA_kwDOC_bI7s8AAAABIO0OZA", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/common", "name": "common", "color": "c5def5", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "@cuichenx / @ericharper, can you review this one? Thanks!", "Merging this because it's only failing on a problematic test which is being commented out in #9085 " ]

[ { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[ { "login": "maanug-nv", "id": 109391026, "node_id": "U_kgDOBoUssg", "avatar_url": "https://avatars.githubusercontent.com/u/109391026?v=4", "gravatar_id": "", "url": "https://api.github.com/users/maanug-nv", "html_url": "https://github.com/maanug-nv", "followers_url": "https://api.github.com/users/maanug-nv/followers", "following_url": "https://api.github.com/users/maanug-nv/following{/other_user}", "gists_url": "https://api.github.com/users/maanug-nv/gists{/gist_id}", "starred_url": "https://api.github.com/users/maanug-nv/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/maanug-nv/subscriptions", "organizations_url": "https://api.github.com/users/maanug-nv/orgs", "repos_url": "https://api.github.com/users/maanug-nv/repos", "events_url": "https://api.github.com/users/maanug-nv/events{/privacy}", "received_events_url": "https://api.github.com/users/maanug-nv/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[]

[]

[]

[]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" } ]

[]

[]

[ { "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" }, { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null } ]

[]

[ "55 minutes is high for fastconformer models with full attention. \r\nCould you try with this model: https://huggingface.co/nvidia/parakeet-tdt_ctc-1.1b ", "Thank you, will try this out.\r\nBy the way as per the documentation in hugging face under which license does this and other parakeet models (rnnt) and come under? It says cc by 4.0 does this grant it commercial use?\r\n", "yes, cc-by-4.0 grants commercial usage ", "Thank you,\r\nI tried using the parakeet-tdt_ctc-1.1b model, inference time on GPU for long form audio is really good. The WER is also small except for names and surnames but that is understandable.\r\n\r\nI noticed one issue is that the model output had sporadic errors where it does not apply a space between two words randomly. Not sure why This issue occurs, for eg when it is supposed to transcribe 'that is' or 'community driven' it instead outputs thatis or communitydriven.\r\n\r\nIs this an inherent issue in the model output or is there some concatenation code which might be creating this issue, I used the model using the sample code in Huggingface:\r\n\r\nimport nemo.collections.asr as nemo_asr\r\nasr_model = nemo_asr.models.ASRModel.from_pretrained(\"nvidia/parakeet-tdt_ctc-1.1b\")\r\n\r\ntranscriptions = asr_model.transcribe([\"sample.wav\"])\r\n\r\nprint(transcriptions)\r\n\r\nI wanted to know if this was something that can be mitigated before final output is produced or is it inherent to how the model outputs data", "Thanks for your feedback, its wonderful that it solved the longform inference problem. \r\n\r\nIts part of model, and provides PnC by default. Will consider to provide a new version with improvements. ", "I am assuming by PnC you mean Punctuation and Capitalization. if so, then that is not the issue I was highlighting. I meant to say that when the transcribed words should be 'that is' or 'community driven' it removed the spaces between the two words in the output. So the output is 'thatis' or 'communitydriven'\r\n\r\nIs this because of the concatenation method used to concat output strings? and if so can we correct it before final output is generated as the code provided does not include any concat methods only inference method:\r\n\r\nimport nemo.collections.asr as nemo_asr\r\nasr_model = nemo_asr.models.ASRModel.from_pretrained(\"nvidia/parakeet-tdt_ctc-1.1b\")\r\n\r\ntranscriptions = asr_model.transcribe([\"sample.wav\"])\r\n\r\nprint(transcriptions)\r\n\r\n", "This issue is stale because it has been open for 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.", "This issue was closed because it has been inactive for 7 days since being marked as stale." ]

[ { "id": 4811886210, "node_id": "LA_kwDOC_bI7s8AAAABHs-Ogg", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/NLP", "name": "NLP", "color": "CBF09A", "default": false, "description": "" }, { "id": 6627254865, "node_id": "LA_kwDOC_bI7s8AAAABiwPiUQ", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Multi%20Modal", "name": "Multi Modal", "color": "ededed", "default": false, "description": null }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[]

[ { "id": 4585431311, "node_id": "LA_kwDOC_bI7s8AAAABEVAhDw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/stale", "name": "stale", "color": "ededed", "default": false, "description": null }, { "id": 4840216855, "node_id": "LA_kwDOC_bI7s8AAAABIH_ZFw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/CI", "name": "CI", "color": "1D76DB", "default": false, "description": "" }, { "id": 6876332615, "node_id": "LA_kwDOC_bI7s8AAAABmdyCRw", "url": "https://api.github.com/repos/NVIDIA/NeMo/labels/Run%20CICD", "name": "Run CICD", "color": "fef2c0", "default": false, "description": "" } ]

[]

[ "This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days.", "This PR was closed because it has been inactive for 7 days since being marked as stale." ]