{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":130871393,"defaultBranch":"master","name":"distiller","ownerLogin":"IntelLabs","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2018-04-24T14:58:12.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/1492758?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1682351572.0","currentOid":""},"activityList":{"items":[{"before":"9e895397954a5bec00c95a1bbde81d8e5bbfbce1","after":"16ea1bb3b68d99bb4ce6e767c5c1e96c9c717caf","ref":"refs/heads/4-bit-qat","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"add arbitrary value for 8 bit gaussian clipping to avoid crushing","shortMessageHtmlLink":"add arbitrary value for 8 bit gaussian clipping to avoid crushing"}},{"before":"f3c8fbac0710c737cb096d02dd32689aa24f4b93","after":"555f88ab327385df11501676b6ac9e35f37d11c6","ref":"refs/heads/scale_approximation","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"ignore dropout during stat collection","shortMessageHtmlLink":"ignore dropout during stat collection"}},{"before":"42035b75eff1aa54ff64ec09b19283e2010ce576","after":"7e8898d537db672862d3c76d0548fc62b5679ac5","ref":"refs/heads/ptq_propagate_settings","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Bug fix in PTQ yaml","shortMessageHtmlLink":"Bug fix in PTQ yaml"}},{"before":"dda6a5ac49f57f7b2651106c19401c858e5e1132","after":"85c449a05091a74f12a1ee7dc92d4b60e7435ad4","ref":"refs/heads/lattice_regularization","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":".","shortMessageHtmlLink":"."}},{"before":"01cb0cd612c77b63fb306dc23818ef1a5cf7eac8","after":"8cdfee44db9459e0d5d0c38e83dcdc07d9c0a032","ref":"refs/heads/update_links","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Fix few more broken links","shortMessageHtmlLink":"Fix few more broken links"}},{"before":"cdea37c2ce770ba1a28c55617f4a06cee442b3b9","after":"b7d1ff6cdeb65746f0a3faec0a326e5ec4c04c88","ref":"refs/heads/load_quant_aware_train_checkpoint","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Merge branch 'master' into load_quant_aware_train_checkpoint","shortMessageHtmlLink":"Merge branch 'master' into load_quant_aware_train_checkpoint"}},{"before":"53275ce4868b5816530fc18d0077b638f549eefd","after":"c7224453a6b2d4c1623e4c51cc1bbb7f0a7ff250","ref":"refs/heads/kd_fixes","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Move + rename shell script","shortMessageHtmlLink":"Move + rename shell script"}},{"before":"98cd087020e75544c7696cdcece745e6a6d44aff","after":"d859fc85a3c9c3048442a4ccd9b98ac45f1e0dc4","ref":"refs/heads/master","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"removing references to aws S3 bucket","shortMessageHtmlLink":"removing references to aws S3 bucket"}},{"before":"f3dd6f19d73237d42649ca2bdf3792cc8b2d4bb8","after":"dcee87a2c5aea6328cabe7ed5e627b71ba14d432","ref":"refs/heads/ax_integration","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Initial example of integration of FB's Ax library","shortMessageHtmlLink":"Initial example of integration of FB's Ax library"}},{"before":"20044dfdf83af3aec13f9fe0b462ea8a3f7bf762","after":"e965897f4f7bd20bec6fba5985f062784bfacaa8","ref":"refs/heads/api_changes","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"small tensor masking API refactoring\n\nAdded masking primitives:\n -mask_tensor\n -create_mask_threshold_criterion\n -create_mask_level_criterion\n -create_mask_sensitivity_criterion\n\n These APIs have a clearer name and communicate their\n responsibility better: create a tensor mask, based on\n some criterion.  Previously,\n distiller.pruning.create_mask_threshold_criterion was\n named distiller.threshold_mask which did not communicate\n well what this function did.\n Masking functionality is no longer hidden\n inside the Pruner instances, so they can be used directly\n by an application, or to compose new Pruner classes.\n\nRemoved file distiller.pruning.pruner:\n -The base-class _ParameterPruner is useless and adds\n needless details to the implementation.\n\nAGP: Separated the pruning-rate schedule from the\n rest of the logic.  This allows us to mix-and-match different\n pruning-rate schedules (just like LR schedulers).","shortMessageHtmlLink":"small tensor masking API refactoring"}},{"before":"6a4fec3d568776e8c773ee7ac0d285f72ab531fd","after":"f5ad5dfb38e151f35e7228bf7f973a0f0046f9cb","ref":"refs/heads/gste","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"lsq for weights","shortMessageHtmlLink":"lsq for weights"}},{"before":"16d722f3b7bbb8b551b4d486be378f0850204e93","after":"ff3da0fa2518ac8907c432e327ecf9a94385dd3e","ref":"refs/heads/inception_support","pushedAt":"2023-04-24T16:11:47.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Special handling for inception+googlenet pre-processing","shortMessageHtmlLink":"Special handling for inception+googlenet pre-processing"}},{"before":"9cf20b2aa87371f22d0f3fae7bfb2118d88fa455","after":"01cb0cd612c77b63fb306dc23818ef1a5cf7eac8","ref":"refs/heads/update_links","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Fix few more broken links","shortMessageHtmlLink":"Fix few more broken links"}},{"before":"b38f45df80016a80cb30a41a626f6121306dd5fe","after":"16d722f3b7bbb8b551b4d486be378f0850204e93","ref":"refs/heads/inception_support","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Special handling for inception+googlenet pre-processing","shortMessageHtmlLink":"Special handling for inception+googlenet pre-processing"}},{"before":"6c310fefed8d504c925c3db4c6c33d6425e53bbc","after":"20044dfdf83af3aec13f9fe0b462ea8a3f7bf762","ref":"refs/heads/api_changes","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"small tensor masking API refactoring\n\nAdded masking primitives:\n -mask_tensor\n -create_mask_threshold_criterion\n -create_mask_level_criterion\n -create_mask_sensitivity_criterion\n\n These APIs have a clearer name and communicate their\n responsibility better: create a tensor mask, based on\n some criterion.  Previously,\n distiller.pruning.create_mask_threshold_criterion was\n named distiller.threshold_mask which did not communicate\n well what this function did.\n Masking functionality is no longer hidden\n inside the Pruner instances, so they can be used directly\n by an application, or to compose new Pruner classes.\n\nRemoved file distiller.pruning.pruner:\n -The base-class _ParameterPruner is useless and adds\n needless details to the implementation.\n\nAGP: Separated the pruning-rate schedule from the\n rest of the logic.  This allows us to mix-and-match different\n pruning-rate schedules (just like LR schedulers).","shortMessageHtmlLink":"small tensor masking API refactoring"}},{"before":"73f9e93fb27cd77be167bff5aa3e139ed3e77104","after":"42035b75eff1aa54ff64ec09b19283e2010ce576","ref":"refs/heads/ptq_propagate_settings","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Bug fix in PTQ yaml","shortMessageHtmlLink":"Bug fix in PTQ yaml"}},{"before":"c5f5b35ac38a94006514bc29f6657d0d80305544","after":"53275ce4868b5816530fc18d0077b638f549eefd","ref":"refs/heads/kd_fixes","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Move + rename shell script","shortMessageHtmlLink":"Move + rename shell script"}},{"before":"9b4c80474191868677a5d326d2e9bd468998fa46","after":"6a4fec3d568776e8c773ee7ac0d285f72ab531fd","ref":"refs/heads/gste","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"lsq for weights","shortMessageHtmlLink":"lsq for weights"}},{"before":"e5c368d10be92f4f7306c520a8502d93b328300d","after":"98cd087020e75544c7696cdcece745e6a6d44aff","ref":"refs/heads/master","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"removing references to aws S3 bucket","shortMessageHtmlLink":"removing references to aws S3 bucket"}},{"before":"e91e07cf8e7eed96e0c9aea584bc09527cf53866","after":"9e895397954a5bec00c95a1bbde81d8e5bbfbce1","ref":"refs/heads/4-bit-qat","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"add arbitrary value for 8 bit gaussian clipping to avoid crushing","shortMessageHtmlLink":"add arbitrary value for 8 bit gaussian clipping to avoid crushing"}},{"before":"c135dd052e5c8f0aede55682dacac225d93eba56","after":"cdea37c2ce770ba1a28c55617f4a06cee442b3b9","ref":"refs/heads/load_quant_aware_train_checkpoint","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Merge branch 'master' into load_quant_aware_train_checkpoint","shortMessageHtmlLink":"Merge branch 'master' into load_quant_aware_train_checkpoint"}},{"before":"59115753519a5ead56c62ead63eef9bf9c6a4ccb","after":"dda6a5ac49f57f7b2651106c19401c858e5e1132","ref":"refs/heads/lattice_regularization","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":".","shortMessageHtmlLink":"."}},{"before":"959ee0c71cf9f2bd8d2f566ec04565e39c8295ab","after":"f3c8fbac0710c737cb096d02dd32689aa24f4b93","ref":"refs/heads/scale_approximation","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"ignore dropout during stat collection","shortMessageHtmlLink":"ignore dropout during stat collection"}},{"before":"5701cbdab510b827ca38ebe583fdef75a4fa2a47","after":"f3dd6f19d73237d42649ca2bdf3792cc8b2d4bb8","ref":"refs/heads/ax_integration","pushedAt":"2023-04-24T16:09:40.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Initial example of integration of FB's Ax library","shortMessageHtmlLink":"Initial example of integration of FB's Ax library"}},{"before":"9e1fcef1aac10a21c071716ac89f2799db6ee243","after":"5701cbdab510b827ca38ebe583fdef75a4fa2a47","ref":"refs/heads/ax_integration","pushedAt":"2023-04-24T15:56:25.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Initial example of integration of FB's Ax library","shortMessageHtmlLink":"Initial example of integration of FB's Ax library"}},{"before":"2c9202d51ea8965a2442fea4d898afb731af518f","after":"59115753519a5ead56c62ead63eef9bf9c6a4ccb","ref":"refs/heads/lattice_regularization","pushedAt":"2023-04-24T15:56:25.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":".","shortMessageHtmlLink":"."}},{"before":"8be2e56e0f4cc57f93ace3daff96e3495777629a","after":"6c310fefed8d504c925c3db4c6c33d6425e53bbc","ref":"refs/heads/api_changes","pushedAt":"2023-04-24T15:56:25.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"small tensor masking API refactoring\n\nAdded masking primitives:\n -mask_tensor\n -create_mask_threshold_criterion\n -create_mask_level_criterion\n -create_mask_sensitivity_criterion\n\n These APIs have a clearer name and communicate their\n responsibility better: create a tensor mask, based on\n some criterion.  Previously,\n distiller.pruning.create_mask_threshold_criterion was\n named distiller.threshold_mask which did not communicate\n well what this function did.\n Masking functionality is no longer hidden\n inside the Pruner instances, so they can be used directly\n by an application, or to compose new Pruner classes.\n\nRemoved file distiller.pruning.pruner:\n -The base-class _ParameterPruner is useless and adds\n needless details to the implementation.\n\nAGP: Separated the pruning-rate schedule from the\n rest of the logic.  This allows us to mix-and-match different\n pruning-rate schedules (just like LR schedulers).","shortMessageHtmlLink":"small tensor masking API refactoring"}},{"before":"0caca675d322362a96d18b436228466d790cd945","after":"73f9e93fb27cd77be167bff5aa3e139ed3e77104","ref":"refs/heads/ptq_propagate_settings","pushedAt":"2023-04-24T15:56:25.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"Bug fix in PTQ yaml","shortMessageHtmlLink":"Bug fix in PTQ yaml"}},{"before":"75ea63673ef3bc9491e320d48dd329930a15d86a","after":"959ee0c71cf9f2bd8d2f566ec04565e39c8295ab","ref":"refs/heads/scale_approximation","pushedAt":"2023-04-24T15:56:25.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"ignore dropout during stat collection","shortMessageHtmlLink":"ignore dropout during stat collection"}},{"before":"e9af8761937acdc00c39812c20eb2d94d8ad6896","after":"9b4c80474191868677a5d326d2e9bd468998fa46","ref":"refs/heads/gste","pushedAt":"2023-04-24T15:56:25.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"michaelbeale-IL","name":"Michael Beale","path":"/michaelbeale-IL","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/63321611?s=80&v=4"},"commit":{"message":"lsq for weights","shortMessageHtmlLink":"lsq for weights"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAADHyzi6QA","startCursor":null,"endCursor":null}},"title":"Activity · IntelLabs/distiller"}