{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":338456109,"defaultBranch":"main","name":"d2go","ownerLogin":"facebookresearch","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2021-02-12T23:19:11.000Z","ownerAvatar":"https://github.com/avatars/u/16943930?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1717709335.0","currentOid":""},"activityList":{"items":[{"before":"938d372ef2c800e6dbeee588e9f38dd4c75af281","after":"e09224b80d7691fecb54705135c3fde23026d6ed","ref":"refs/heads/main","pushedAt":"2024-07-30T18:52:10.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"remove from pytorch-nightly build in CI\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/672\n\nNightly is less stable than d2go itself (eg. issue like https://github.com/facebookresearch/d2go/actions/runs/10135977974), just use the latest build.\n\nDifferential Revision: D60458684\n\nfbshipit-source-id: 2cce9a0eaabdeba2908703753d67dcd4bb24c378","shortMessageHtmlLink":"remove from pytorch-nightly build in CI"}},{"before":"e6d115e997ae1f5bb1dabe89febe18fc6c5795f0","after":"938d372ef2c800e6dbeee588e9f38dd4c75af281","ref":"refs/heads/main","pushedAt":"2024-07-30T16:47:56.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"fix test_integration_tests_coco\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/671\n\n> AttributeError: type object 'GeneralizedRCNNTask' has no attribute 'cleanup'\n\nEZ fix\n\nReviewed By: rbasch\n\nDifferential Revision: D60400187\n\nfbshipit-source-id: 25872f4928cf8851ff63e96311c12086e272d619","shortMessageHtmlLink":"fix test_integration_tests_coco"}},{"before":"55e0dae3b82579a97928e94f80137f0e10b92cf1","after":"e6d115e997ae1f5bb1dabe89febe18fc6c5795f0","ref":"refs/heads/main","pushedAt":"2024-07-11T16:17:59.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"added an arch without trunk3 for code detection\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/670\n\nregistered a new arch without trunk3 since our trunk2 model has better quality.\n\nReviewed By: huiyujie\n\nDifferential Revision: D59613942\n\nfbshipit-source-id: 605e8925bfcd91d8a966303d9c0a3b4f56a9a0c7","shortMessageHtmlLink":"added an arch without trunk3 for code detection"}},{"before":"040a7167f047ffe6dbdecaea85e7fdff4db611c7","after":"55e0dae3b82579a97928e94f80137f0e10b92cf1","ref":"refs/heads/main","pushedAt":"2024-07-01T19:17:03.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Use different signpost id for each checkpoint\n\nSummary:\nUse different signpost so they don't get deduplicated in minion.\n\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/669\n\nReviewed By: wat3rBro\n\nDifferential Revision: D59226344\n\nfbshipit-source-id: c1356feadbc1b63220a1abdd8cc079723b230e42","shortMessageHtmlLink":"Use different signpost id for each checkpoint"}},{"before":"8eab506b12436bb354be9f28820bdde913127fbb","after":"040a7167f047ffe6dbdecaea85e7fdff4db611c7","ref":"refs/heads/main","pushedAt":"2024-06-22T00:19:54.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"adhere to lazy import rules\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/668\n\nLazy import changes `Python` import semantics, specifically when it comes to initialization of packages/modules: https://www.internalfb.com/intern/wiki/Python/Cinder/Onboarding/Tutorial/Lazy_Imports/Troubleshooting/\n\nFor example, this pattern is not guaranteed to work:\n\n```\nimport torch.optim\n...\ntorch.optim._multi_tensor.Adam   # may fail to resolve _multi_tensor\n```\n\nAnd this is guaranteed to work:\n\n```\nimport torch.optim._multi_tensor\n...\ntorch.optim._multi_tensor.Adam   # will always work\n```\n\nA recent change to `PyTorch` changed module initialization logic in a way that exposed this issue.\n\nBut the code has been working for years? This is the nature of undefined behavior, any change in the environment (in this the `PyTorch` code base can make it fail.\n\nReviewed By: wat3rBro\n\nDifferential Revision: D58876582\n\nfbshipit-source-id: c8f3f53605822517d646e57ddbf4359af54dba0d","shortMessageHtmlLink":"adhere to lazy import rules"}},{"before":"2005474802109693e03348c4e64bbc394568095f","after":"8eab506b12436bb354be9f28820bdde913127fbb","ref":"refs/heads/main","pushedAt":"2024-06-19T08:27:43.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Add preemption checkpointing to lightning tasks\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/666\n\nWhile debugging elevated preemption wastage in d2go, came across a few long running Pinocchio jobs in d2go that do not checkpoint preemption and also do not have checkpointing instrumented. This diff addresses both of these issues.\n\nReviewed By: wat3rBro\n\nDifferential Revision: D58669254\n\nfbshipit-source-id: 9d1c5ff9e61a4a83d284a45154aa54d2d41178cf","shortMessageHtmlLink":"Add preemption checkpointing to lightning tasks"}},{"before":"c606fdf4731a0b447c87e311d6db057862cfb768","after":"2005474802109693e03348c4e64bbc394568095f","ref":"refs/heads/main","pushedAt":"2024-06-11T07:06:28.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Custom qconfig for eager-mode quantization\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/664\n\nGenerate qconfig based on the config\n\nReviewed By: wat3rBro\n\nDifferential Revision: D58210321\n\nfbshipit-source-id: 7a86f8b6e9d112302c978080c2bd5721e3c7dbff","shortMessageHtmlLink":"Custom qconfig for eager-mode quantization"}},{"before":"45b29a0a927d22e3d2feab02230f169e8605ba8f","after":"dc4c33bdf1f14ea3505bf87228c36b72b1ee4f65","ref":"refs/heads/export-D58210321","pushedAt":"2024-06-10T20:28:27.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"navsud","name":"Naveen Suda","path":"/navsud","primaryAvatarUrl":"https://github.com/avatars/u/99509021?s=80&v=4"},"commit":{"message":"Custom qconfig for eager-mode quantization\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/664\n\nGenerate qconfig based on the config\n\nReviewed By: wat3rBro\n\nDifferential Revision: D58210321","shortMessageHtmlLink":"Custom qconfig for eager-mode quantization"}},{"before":"fee05c722f6892099b69189da7e3db799e6257dc","after":"45b29a0a927d22e3d2feab02230f169e8605ba8f","ref":"refs/heads/export-D58210321","pushedAt":"2024-06-08T23:48:11.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"navsud","name":"Naveen Suda","path":"/navsud","primaryAvatarUrl":"https://github.com/avatars/u/99509021?s=80&v=4"},"commit":{"message":"Custom qconfig for eager-mode quantization\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/664\n\nGenerate qconfig based on the config\n\nReviewed By: wat3rBro\n\nDifferential Revision: D58210321","shortMessageHtmlLink":"Custom qconfig for eager-mode quantization"}},{"before":"294cdaf8c1c0713189161e7f4bae0f9ce84a9754","after":"fee05c722f6892099b69189da7e3db799e6257dc","ref":"refs/heads/export-D58210321","pushedAt":"2024-06-08T00:13:12.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Custom qconfig for eager-mode quantization\n\nSummary:\n\nGenerate qconfig based on the config\n\nReviewed By: wat3rBro\n\nDifferential Revision: D58210321","shortMessageHtmlLink":"Custom qconfig for eager-mode quantization"}},{"before":null,"after":"294cdaf8c1c0713189161e7f4bae0f9ce84a9754","ref":"refs/heads/export-D58210321","pushedAt":"2024-06-06T21:28:55.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Custom qconfig for eager-mode quantization\n\nSummary: Generate qconfig based on the config\n\nDifferential Revision: D58210321","shortMessageHtmlLink":"Custom qconfig for eager-mode quantization"}},{"before":"aa7716be479f9d93cbf9c9d0a804f0444ece9bce","after":"c606fdf4731a0b447c87e311d6db057862cfb768","ref":"refs/heads/main","pushedAt":"2024-05-08T05:21:53.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"small fixes of github CI\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/662\n\n- d2go: aws model zoo is not available anymore, disable the test in oss.\n- d2: scripts running at midnight hitting rate limit issue, change it to a random time.\n\nDifferential Revision: D57085427\n\nfbshipit-source-id: 8dc24b2a7996c8ae5ed8808c3301af2851c15a14","shortMessageHtmlLink":"small fixes of github CI"}},{"before":"05b330182a2ff39df5231eac7606f39e24a7b45c","after":"aa7716be479f9d93cbf9c9d0a804f0444ece9bce","ref":"refs/heads/main","pushedAt":"2024-05-02T18:25:51.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Make lightning reproducible\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/661\n\nX-link: https://github.com/fairinternal/detectron2/pull/603\n\nX-link: https://github.com/facebookresearch/detectron2/pull/5273\n\nIn this diff we make changes to ensure we can control reproducibility in d2go:\n\n- update setup.py to enforce deterministic performance if set via config\n- set lightning parameters if deterministic is passed:\n\n```\n {\n                \"sync_batchnorm\": True,\n                \"deterministic\": True,\n                \"replace_sampler_ddp\": False,\n }\n```\n- allow passing prefetch_factor, pin_memory, persistent_memory as args to batch dataloader.\n- minor fix in training sampler\n\nDifferential Revision: D55767128\n\nfbshipit-source-id: eeab50c95969a91c58f1773473b6fc666494cc16","shortMessageHtmlLink":"Make lightning reproducible"}},{"before":"ba7c235b77f9f4e4c4a94cbcc56825e285d006f0","after":"05b330182a2ff39df5231eac7606f39e24a7b45c","ref":"refs/heads/main","pushedAt":"2024-04-24T05:02:49.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"use custom prepare function\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/660\n\nTo enable w8a16 sigmoid in d2go, we need to use custom prepare function.\n\nReviewed By: ayushidalmia, jiaxuzhu92\n\nDifferential Revision: D56275899\n\nfbshipit-source-id: 654900011a1393e81289e8c9412b5886831765e2","shortMessageHtmlLink":"use custom prepare function"}},{"before":"dda734a452031029fedb8788f734cd76329aa624","after":"ba7c235b77f9f4e4c4a94cbcc56825e285d006f0","ref":"refs/heads/main","pushedAt":"2024-04-03T17:40:34.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Add new FBNetV3_B_large backbone for HAPTIC.\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/659\n\nFBNetV3_B_large using FBNetV3_B backbone with large box and kpts heads for better multitask KP+Box detection tasks while being in similar range of #flops and #params.\n\nReviewed By: ashishvshenoy\n\nDifferential Revision: D55645349\n\nfbshipit-source-id: 3fe84f566b3eeaddf84a94ef708557944fffcd22","shortMessageHtmlLink":"Add new FBNetV3_B_large backbone for HAPTIC."}},{"before":"b14282fb8380798ebd756781cd3e20a93538681c","after":"dda734a452031029fedb8788f734cd76329aa624","ref":"refs/heads/main","pushedAt":"2024-04-02T21:40:19.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"INPUT.CROP=ENABLED silently fails when annotations dict is present but empty.\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/658\n\nD2Go's INPUT.CROP=ENABLED silently fails when annotations dict is present but empty.\nEnds up not using the dataset at all.\n\nAdding an additional check to circumvent this failure.\n\nReviewed By: ayushidalmia\n\nDifferential Revision: D55640142\n\nfbshipit-source-id: b733b841edb17c16d69332795c89c32b008cb6e5","shortMessageHtmlLink":"INPUT.CROP=ENABLED silently fails when annotations dict is present bu…"}},{"before":"abdad994b73c3afbf92b584598cca12a248ddb09","after":"b14282fb8380798ebd756781cd3e20a93538681c","ref":"refs/heads/main","pushedAt":"2024-03-27T02:54:06.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"fix distributed initialization for FSDP\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/657\n\nWithout properly set `requires_grad` for params and buffers, it causes hang in FSDP training. This becomes an issue eg when training with LoRA.\n\nReviewed By: wat3rBro\n\nDifferential Revision: D55220828\n\nfbshipit-source-id: 1e33aa540c84c4de62a3a37c48a322aa26c98292","shortMessageHtmlLink":"fix distributed initialization for FSDP"}},{"before":"102305a5e9a16f1a6c2fa201660d75cf9e0d9139","after":"abdad994b73c3afbf92b584598cca12a248ddb09","ref":"refs/heads/main","pushedAt":"2024-03-19T20:16:22.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"distributed FSDP model initialization\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/656\n\nEnable distributed FSDP model initialization. This iteratively moves and shards the model on GPU to allow for the training of models greater than single GPU HBM capacity and which cannot be instantiated multiple times on a single host.\n\nThe flow is as follows:\n1. Rank 0 will init the whole model on CPU using existing code paths, while all other ranks init an 'empty' model using fake tensors.\n2. Once this is complete and initialization moves to FSDP, distributed init will traverse the model 'bottom-up', transferring all params/buffers from rank 0 to all other ranks, while simultaneously wrapping modules in FSDP whenever possible (based on the specified config). Thus modules are sharded (and memory usage distributed) at the first possible instance using the existing FSDP api/implementation.\n\nReviewed By: XiaoliangDai\n\nDifferential Revision: D54287718\n\nfbshipit-source-id: 16d63d78065d1fca0c6baf7a385f666a4e1b2a5f","shortMessageHtmlLink":"distributed FSDP model initialization"}},{"before":"1216c22534ca91cb9b33e0aa92b32d9509e3c012","after":"102305a5e9a16f1a6c2fa201660d75cf9e0d9139","ref":"refs/heads/main","pushedAt":"2024-03-14T05:56:23.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"add additional args to custom_prepare_pt2e\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/654\n\nexample_inputs and is_qat args are needed for some models during prepare_pt2e step.\n\nReviewed By: chakriu, tarun292\n\nDifferential Revision: D54873270\n\nfbshipit-source-id: 67df457aca82fd9da77969133ecf390cdc80fb85","shortMessageHtmlLink":"add additional args to custom_prepare_pt2e"}},{"before":"2256bdb7420218f9c435be525adfef0451a3a2f6","after":"1216c22534ca91cb9b33e0aa92b32d9509e3c012","ref":"refs/heads/main","pushedAt":"2024-03-10T08:33:30.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"ensure metadata thing_classes consistency with multiple datasets and category filtering\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/653\n\n# Changes\nIn Mask2Former RC4 training, we need to use a particular weighted category training sampler where `DATALOADER.SAMPLER_TRAIN = \"WeightedCategoryTrainingSampler\"`.\n\nAlso there are multiple datasets are used, and the set of each one's categories are not exactly identical. Some datasets have more categories (e.g. Exo-body) than other datasets that do not have exobody annotations.\n\nAlso we use category filtering by setting `D2GO_DATA.DATASETS.TRAIN_CATEGORIES` to a subset of full categories.\n\nIn this setup, currently D2GO will complain metadata.thing_classes is NOT consistency across datasets (https://fburl.com/code/k8xbvyfd).\n\nThe reason is when category filtering is used, D2GO writes a temporary dataset json file (https://fburl.com/code/slb5z6mc).\nAnd this tmp json file will be loaded when we get the dataset dicts from DatasetCatalog (https://fburl.com/code/5k4ynyhc). Meanwhile, metadata in MetadataCatalog for this category-filtered dataset is also updated based on categories stored in this tmp file.\n\nTherefore, we must ensure categories stored in the tmp file is consistent between multiple category-filtered datasets.\n\nIn this diff, we update the logic of writing such tmp dataset json file.\n\n# Github CI test\nNote **CI / python-unittest-cpu** is shown as failed with error below. But I do not think it is related to changes in this diff since error is related to observer in the QAT model training, but changes in the diff are related to dataset preparation.\n\n```\nTraceback (most recent call last):\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/detectron2/engine/train_loop.py\", line 155, in train\n    self.run_step()\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/detectron2/engine/train_loop.py\", line 310, in run_step\n    loss_dict = self.model(data)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/nn/modules/module.py\", line 1527, in _wrapped_call_impl\n    return self._call_impl(*args, **kwargs)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/nn/modules/module.py\", line 1536, in _call_impl\n    return forward_call(*args, **kwargs)\n  File \"/home/runner/work/d2go/d2go/tests/runner/test_runner_default_runner.py\", line 44, in forward\n    ret = self.conv(images.tensor)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/nn/modules/module.py\", line 1527, in _wrapped_call_impl\n    return self._call_impl(*args, **kwargs)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/nn/modules/module.py\", line 1590, in _call_impl\n    hook_result = hook(self, args, result)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/ao/quantization/quantize.py\", line 131, in _observer_forward_hook\n    return self.activation_post_process(output)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/nn/modules/module.py\", line 1527, in _wrapped_call_impl\n    return self._call_impl(*args, **kwargs)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/nn/modules/module.py\", line 1536, in _call_impl\n    return forward_call(*args, **kwargs)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/ao/quantization/fake_quantize.py\", line 199, in forward\n    _scale, _zero_point = self.calculate_qparams()\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/ao/quantization/fake_quantize.py\", line 194, in calculate_qparams\n    return self.activation_post_process.calculate_qparams()\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/ao/quantization/observer.py\", line 529, in calculate_qparams\n    return self._calculate_qparams(self.min_val, self.max_val)\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/ao/quantization/observer.py\", line 328, in _calculate_qparams\n    if not check_min_max_valid(min_val, max_val):\n  File \"/usr/share/miniconda/envs/__setup_conda/lib/python3.8/site-packages/torch/ao/quantization/utils.py\", line 346, in check_min_max_valid\n    assert min_val <= max_val, f\"min {min_val} should be less than max {max_val}\"\nAssertionError: min 3.8139522075653076e-05 should be less than max -3.8139522075653076e-05\n```\n\nReviewed By: ayushidalmia\n\nDifferential Revision:\nD54665936\n\nPrivacy Context Container: L1243674\n\nfbshipit-source-id: 322ab4a84a710b03fa39b39fa81117752d369ba5","shortMessageHtmlLink":"ensure metadata thing_classes consistency with multiple datasets and …"}},{"before":"09bd2869c3d9d7c655083fffb2652b29ba63e951","after":"2256bdb7420218f9c435be525adfef0451a3a2f6","ref":"refs/heads/main","pushedAt":"2024-03-03T01:45:57.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"apply Black 2024 style in fbcode (7/16)\n\nSummary:\nFormats the covered files with pyfmt.\n\npaintitblack\n\nReviewed By: aleivag\n\nDifferential Revision: D54447732\n\nfbshipit-source-id: e21fbbe27882c8af183d021f4ac27029cbe93e8e","shortMessageHtmlLink":"apply Black 2024 style in fbcode (7/16)"}},{"before":"a637c6cc9ded61313672ab97ebf48787b4965184","after":"09bd2869c3d9d7c655083fffb2652b29ba63e951","ref":"refs/heads/main","pushedAt":"2024-02-23T22:28:03.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"pt2e quantization support in D2Go\n\nSummary: Add pt2e quantization support in D2Go.\n\nReviewed By: chakriu\n\nDifferential Revision: D54132092\n\nfbshipit-source-id: 34a9ba79a5eb49ed27a3f33454078b0df37cf2f0","shortMessageHtmlLink":"pt2e quantization support in D2Go"}},{"before":"2653e2262386d4633cba0223105a6dd4dd6a9f06","after":"a637c6cc9ded61313672ab97ebf48787b4965184","ref":"refs/heads/main","pushedAt":"2024-02-17T02:44:26.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"fix more typo\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/631\n\nomm -> oom\n\nReviewed By: EugenHotaj, wat3rBro\n\nDifferential Revision: D50860125\n\nfbshipit-source-id: 553220106aed1c8c752347a7a5c01b525ec25588","shortMessageHtmlLink":"fix more typo"}},{"before":"d6b2d86058945c1d650ff31020a766841ec390fa","after":"2653e2262386d4633cba0223105a6dd4dd6a9f06","ref":"refs/heads/main","pushedAt":"2024-02-08T01:23:07.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Pass cfg.SEED to dataloader building\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/651\n\nAs title, so that dataloading is deterministic when `cfg.SEED` is set.\n\nReviewed By: navsud\n\nDifferential Revision: D53547772\n\nfbshipit-source-id: 73cfd2b351e81b370fb721a4f7b7c2a6313470bd","shortMessageHtmlLink":"Pass cfg.SEED to dataloader building"}},{"before":null,"after":"d670a270d02b711edbc386bfd37ec7080df7c0e6","ref":"refs/heads/export-D53547772","pushedAt":"2024-02-07T23:13:49.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"Pass cfg.SEED to dataloader building\n\nSummary: As title, so that dataloading is deterministic when `cfg.SEED` is set.\n\nDifferential Revision: D53547772","shortMessageHtmlLink":"Pass cfg.SEED to dataloader building"}},{"before":"3c6f71b4ce8a27da8d39e9746f1f4b75367b2693","after":"d6b2d86058945c1d650ff31020a766841ec390fa","ref":"refs/heads/main","pushedAt":"2024-02-04T19:05:25.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"bfloat16 training\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/649\n\nsupport bfloat16 training\n\nReviewed By: chihyaoma, Sekunde\n\nDifferential Revision: D53029989\n\nfbshipit-source-id: 2e1d8f2112d238441e3f6801db3092383147fdbd","shortMessageHtmlLink":"bfloat16 training"}},{"before":"92af450bdc45adc32d11cef216d839f9bbc5aeb0","after":"3c6f71b4ce8a27da8d39e9746f1f4b75367b2693","ref":"refs/heads/main","pushedAt":"2024-01-17T18:19:33.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"expose example_input argument in setup_qat_model()\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/647\n\nMajor changes\n- **example_input** argument in **prepare_fake_quant_model()** is useful in certain cases. For example, in Argos model **custom_prepare_fx()** method under FX graph + QAT setup (D52760682), it is used to prepare example inputs to individual sub-modules by running one forward pass and bookkeeping the inputs to individual sub-modules. Therefore, we export argument **example_input** in **setup_qat_model()** function.\n- For QAT model, currently we assert # of state dict keys (excluding observers) should be equal to # of state dict keys in the original model. However, when the assertion fails, it does not log useful information for debugging. We make changes to report what are the unique keys in each state dict.\n\nReviewed By: navsud\n\nDifferential Revision: D52760688\n\nfbshipit-source-id: 27535a0324ebe6513f198acb839918a0346720d0","shortMessageHtmlLink":"expose example_input argument in setup_qat_model()"}},{"before":"573bd454c135a9b950943eec33d43380b41c64f2","after":"92af450bdc45adc32d11cef216d839f9bbc5aeb0","ref":"refs/heads/main","pushedAt":"2024-01-16T16:47:40.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"fbcode//mobile-vision/d2go/tests\n\nSummary: Pull Request resolved: https://github.com/facebookresearch/d2go/pull/648\n\nDifferential Revision: D52792992\n\nfbshipit-source-id: d3a64f3b306ea024ec072eaa6327446a84e2d83c","shortMessageHtmlLink":"fbcode//mobile-vision/d2go/tests"}},{"before":"94cf50689eb996717176badeb3ab2af619e25164","after":"573bd454c135a9b950943eec33d43380b41c64f2","ref":"refs/heads/main","pushedAt":"2024-01-12T00:56:32.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"consolidate deterministic settings\n\nSummary:\nPull Request resolved: https://github.com/facebookresearch/d2go/pull/644\n\nThis diff consolidates deterministic settings in D2Go. In the `default_runner.py` file, the `torch.set_float32_matmul_precision(\"highest\")` function is added to set the precision for matrix multiplication to the highest possible value. In the `setup.py` file, the `torch.backends.cudnn.deterministic` setting is set to `True` and the `torch.backends.cudnn.allow_tf32` setting is set to `False` to avoid random pytorch and CUDA algorithms during the training. The `torch.backends.cuda.matmul.allow_tf32` setting is also set to `False` to avoid random matrix multiplication algorithms. Additionally, the `seed` function is used to set the seed for reproducibility.\n\nReviewed By: wat3rBro\n\nDifferential Revision: D51796739\n\nfbshipit-source-id: 50e44ea50b0311b56a885db9f633491ac3002bd4","shortMessageHtmlLink":"consolidate deterministic settings"}},{"before":"cfa858356cd6b45cf64eec8f921879eac63d0edf","after":"94cf50689eb996717176badeb3ab2af619e25164","ref":"refs/heads/main","pushedAt":"2024-01-08T10:37:07.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"facebook-github-bot","name":"Facebook Community Bot","path":"/facebook-github-bot","primaryAvatarUrl":"https://github.com/avatars/u/6422482?s=80&v=4"},"commit":{"message":"mobile-vision/experimental 2/5\n\nSummary: Pull Request resolved: https://github.com/facebookresearch/d2go/pull/646\n\nReviewed By: wat3rBro\n\nDifferential Revision: D52536555\n\nfbshipit-source-id: e57dc5b2774771f0739118c5244014171732c151","shortMessageHtmlLink":"mobile-vision/experimental 2/5"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAEjVIttAA","startCursor":null,"endCursor":null}},"title":"Activity · facebookresearch/d2go"}