{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":235860204,"defaultBranch":"master","name":"DeepSpeed","ownerLogin":"microsoft","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2020-01-23T18:35:18.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/6154722?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1716588253.0","currentOid":""},"activityList":{"items":[{"before":"596de4c68c3ecc5242ab6f9a26f6714dfbfe909b","after":"2ccd27d9754abd468817fda08f16c322367f0195","ref":"refs/heads/duli/cuda_op_builder","pushedAt":"2024-05-24T23:22:46.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"duli2012","name":"Du Li","path":"/duli2012","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2879800?s=80&v=4"},"commit":{"message":"modify import paths","shortMessageHtmlLink":"modify import paths"}},{"before":null,"after":"2007b175640855e073613248b08416874c3e14da","ref":"refs/heads/loadams/fix-website-sidebar","pushedAt":"2024-05-24T22:04:13.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"loadams","name":"Logan Adams","path":"/loadams","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/114770087?s=80&v=4"},"commit":{"message":"Update to fix sidebar over test","shortMessageHtmlLink":"Update to fix sidebar over test"}},{"before":null,"after":"0ad1b8579188c6b072346c3178b2c674e3e92491","ref":"refs/heads/loadams/update-hpu-docker-container","pushedAt":"2024-05-24T20:38:43.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"loadams","name":"Logan Adams","path":"/loadams","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/114770087?s=80&v=4"},"commit":{"message":"Update HPU docker version","shortMessageHtmlLink":"Update HPU docker version"}},{"before":"d80508d5468707e1447329e410622f44cff14371","after":"8c0b98f2bee3a4bcf4cb0c8920928cb2151c586b","ref":"refs/heads/mrwyattii/pydantic-2-support","pushedAt":"2024-05-24T19:21:03.000Z","pushType":"push","commitsCount":3,"pusher":{"login":"adk9","name":"Abhishek Kulkarni","path":"/adk9","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/11399?s=80&v=4"},"commit":{"message":"Merge branch 'master' into mrwyattii/pydantic-2-support","shortMessageHtmlLink":"Merge branch 'master' into mrwyattii/pydantic-2-support"}},{"before":"5e5cc847f3b463fae130ab84f149b62a63197fab","after":"9d176cf431db6eee86632d08cd76f2131dd8099b","ref":"refs/heads/cholmes/comm-group-cache","pushedAt":"2024-05-23T17:53:38.000Z","pushType":"push","commitsCount":306,"pusher":{"login":"jomayeri","name":"Joe Mayer","path":"/jomayeri","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/114769929?s=80&v=4"},"commit":{"message":"Merge branch 'master' into cholmes/comm-group-cache","shortMessageHtmlLink":"Merge branch 'master' into cholmes/comm-group-cache"}},{"before":"3a3a6db3332e339cc9fd94efd4982f6d60635a3d","after":null,"ref":"refs/heads/gh-readonly-queue/master/pr-5562-f86824be81b508a365c36ec0e900e92cce34d8e4","pushedAt":"2024-05-23T16:25:43.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"}},{"before":"f86824be81b508a365c36ec0e900e92cce34d8e4","after":"3a3a6db3332e339cc9fd94efd4982f6d60635a3d","ref":"refs/heads/master","pushedAt":"2024-05-23T16:25:41.000Z","pushType":"merge_queue_merge","commitsCount":1,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow (#5562)\n\nThis PR updates the `nv-ds-chat` GitHub workflow to include\n`hybrid_engine.py` file in the path. This is done to ensure testing on\nthe DS-Chat flow is done whenever any changes are made to the Hybrid\nEngine.","shortMessageHtmlLink":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow (#5562)"}},{"before":null,"after":"3a3a6db3332e339cc9fd94efd4982f6d60635a3d","ref":"refs/heads/gh-readonly-queue/master/pr-5562-f86824be81b508a365c36ec0e900e92cce34d8e4","pushedAt":"2024-05-23T15:21:19.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow (#5562)\n\nThis PR updates the `nv-ds-chat` GitHub workflow to include\n`hybrid_engine.py` file in the path. This is done to ensure testing on\nthe DS-Chat flow is done whenever any changes are made to the Hybrid\nEngine.","shortMessageHtmlLink":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow (#5562)"}},{"before":"0e0c14d94ca5a2865b5e9716e4f3b5f744ff6730","after":null,"ref":"refs/heads/gh-readonly-queue/master/pr-5562-f86824be81b508a365c36ec0e900e92cce34d8e4","pushedAt":"2024-05-23T01:10:17.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"}},{"before":null,"after":"0e0c14d94ca5a2865b5e9716e4f3b5f744ff6730","ref":"refs/heads/gh-readonly-queue/master/pr-5562-f86824be81b508a365c36ec0e900e92cce34d8e4","pushedAt":"2024-05-23T00:16:51.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow (#5562)\n\nThis PR updates the `nv-ds-chat` GitHub workflow to include\n`hybrid_engine.py` file in the path. This is done to ensure testing on\nthe DS-Chat flow is done whenever any changes are made to the Hybrid\nEngine.","shortMessageHtmlLink":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow (#5562)"}},{"before":"9663533e1789bafa16de7956d7f42927385fee58","after":"d252919b08e8961c48fb8f76ece8041de7e7de1a","ref":"refs/heads/lekurile/update_dschat_wf","pushedAt":"2024-05-22T23:11:02.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"lekurile","name":"Lev Kurilenko","path":"/lekurile","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/113481193?s=80&v=4"},"commit":{"message":"Merge branch 'master' into lekurile/update_dschat_wf","shortMessageHtmlLink":"Merge branch 'master' into lekurile/update_dschat_wf"}},{"before":"f86824be81b508a365c36ec0e900e92cce34d8e4","after":null,"ref":"refs/heads/gh-readonly-queue/master/pr-5525-995ba11928f2f264e96f2b80aa5781a826b445fe","pushedAt":"2024-05-22T22:57:05.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"}},{"before":"995ba11928f2f264e96f2b80aa5781a826b445fe","after":"f86824be81b508a365c36ec0e900e92cce34d8e4","ref":"refs/heads/master","pushedAt":"2024-05-22T22:57:04.000Z","pushType":"merge_queue_merge","commitsCount":1,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Add Ulysses DistributedAttention compatibility (#5525)\n\nThe `DistributedAttention` in DeepSpeed-Ulysses has a compatibility with\nthe training code in\n[Megatron-DeepSpeed](https://github.com/microsoft/Megatron-DeepSpeed/blob/main/megatron/model/transformer.py#L811)\nbecause it only takes sequential sequences as input parameters. However,\nthis is not compatible with the frequently used scenarios of specifying\nparameters, such as the following scenario when using Flash Attention:\n```python\nulysses_attn = DistributedAttention(local_attention=flash_attn_func, sequence_process_group=None, scatter_idx=2, gather_idx=1)\n\nattn_output = ulysses_attn(\n query_states,\n key_states,\n value_states,\n dropout,\n softmax_scale,\n causal=causal,\n)\n\n```\nTherefore, the `**kwargs` parameter has been added to increase\ncompatibility with more local attention, while making minimal code\nmodifications.\n\nCo-authored-by: Kwen-Chen <2133949025@qq.com>\nCo-authored-by: Olatunji Ruwase \nCo-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>","shortMessageHtmlLink":"Add Ulysses DistributedAttention compatibility (#5525)"}},{"before":null,"after":"9663533e1789bafa16de7956d7f42927385fee58","ref":"refs/heads/lekurile/update_dschat_wf","pushedAt":"2024-05-22T22:46:00.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"lekurile","name":"Lev Kurilenko","path":"/lekurile","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/113481193?s=80&v=4"},"commit":{"message":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow","shortMessageHtmlLink":"Add hybrid_engine.py as path to trigger the DS-Chat GH workflow"}},{"before":null,"after":"b8faade87973fef5394a51ad67afba16fd2371ba","ref":"refs/heads/lekurile/test_rearrange_ops","pushedAt":"2024-05-22T22:39:55.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"lekurile","name":"Lev Kurilenko","path":"/lekurile","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/113481193?s=80&v=4"},"commit":{"message":"Merge branch 'master' into rearrange_ops","shortMessageHtmlLink":"Merge branch 'master' into rearrange_ops"}},{"before":"fcee6a7e2670be52c624d0864698fc165fd7fcb0","after":"d80508d5468707e1447329e410622f44cff14371","ref":"refs/heads/mrwyattii/pydantic-2-support","pushedAt":"2024-05-22T22:27:18.000Z","pushType":"push","commitsCount":9,"pusher":{"login":"loadams","name":"Logan Adams","path":"/loadams","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/114770087?s=80&v=4"},"commit":{"message":"Merge branch 'master' into mrwyattii/pydantic-2-support","shortMessageHtmlLink":"Merge branch 'master' into mrwyattii/pydantic-2-support"}},{"before":null,"after":"f86824be81b508a365c36ec0e900e92cce34d8e4","ref":"refs/heads/gh-readonly-queue/master/pr-5525-995ba11928f2f264e96f2b80aa5781a826b445fe","pushedAt":"2024-05-22T21:52:55.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Add Ulysses DistributedAttention compatibility (#5525)\n\nThe `DistributedAttention` in DeepSpeed-Ulysses has a compatibility with\nthe training code in\n[Megatron-DeepSpeed](https://github.com/microsoft/Megatron-DeepSpeed/blob/main/megatron/model/transformer.py#L811)\nbecause it only takes sequential sequences as input parameters. However,\nthis is not compatible with the frequently used scenarios of specifying\nparameters, such as the following scenario when using Flash Attention:\n```python\nulysses_attn = DistributedAttention(local_attention=flash_attn_func, sequence_process_group=None, scatter_idx=2, gather_idx=1)\n\nattn_output = ulysses_attn(\n query_states,\n key_states,\n value_states,\n dropout,\n softmax_scale,\n causal=causal,\n)\n\n```\nTherefore, the `**kwargs` parameter has been added to increase\ncompatibility with more local attention, while making minimal code\nmodifications.\n\nCo-authored-by: Kwen-Chen <2133949025@qq.com>\nCo-authored-by: Olatunji Ruwase \nCo-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>","shortMessageHtmlLink":"Add Ulysses DistributedAttention compatibility (#5525)"}},{"before":"995ba11928f2f264e96f2b80aa5781a826b445fe","after":null,"ref":"refs/heads/gh-readonly-queue/master/pr-5363-f4efef21b8370f5963f312a52968045b2f5e00b2","pushedAt":"2024-05-22T21:41:01.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"}},{"before":"f4efef21b8370f5963f312a52968045b2f5e00b2","after":"995ba11928f2f264e96f2b80aa5781a826b445fe","ref":"refs/heads/master","pushedAt":"2024-05-22T21:41:00.000Z","pushType":"merge_queue_merge","commitsCount":1,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Add throughput timer configuration (#5363)\n\nThe new \"timers\" section describes configuration for different timers.\n\nSpecifically, in the \"throughput\" section, it is possible to disable the\nthroughput timer (enabled by default). This allows to avoid the\nperformance degradation whenever the throughput measurement is not\nneeded, for example in production environment.\n\nNo device synchronize() is invoked when \"synchronized\" is set to False\n(default is True). This allows to produce approximate throughput\nmeasurements with minimal performance penalty.\n\n---------\n\nCo-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>\nCo-authored-by: Olatunji Ruwase ","shortMessageHtmlLink":"Add throughput timer configuration (#5363)"}},{"before":"f4efef21b8370f5963f312a52968045b2f5e00b2","after":null,"ref":"refs/heads/gh-readonly-queue/master/pr-5546-263bfe2892c1ee6285076214bb5e3898c35e78f3","pushedAt":"2024-05-22T21:28:12.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"}},{"before":"263bfe2892c1ee6285076214bb5e3898c35e78f3","after":"f4efef21b8370f5963f312a52968045b2f5e00b2","ref":"refs/heads/master","pushedAt":"2024-05-22T21:28:11.000Z","pushType":"merge_queue_merge","commitsCount":1,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"[INF] DSAttention allow input_mask to have false as value (#5546)\n\nCo-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>","shortMessageHtmlLink":"[INF] DSAttention allow input_mask to have false as value (#5546)"}},{"before":null,"after":"995ba11928f2f264e96f2b80aa5781a826b445fe","ref":"refs/heads/gh-readonly-queue/master/pr-5363-f4efef21b8370f5963f312a52968045b2f5e00b2","pushedAt":"2024-05-22T20:28:17.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Add throughput timer configuration (#5363)\n\nThe new \"timers\" section describes configuration for different timers.\n\nSpecifically, in the \"throughput\" section, it is possible to disable the\nthroughput timer (enabled by default). This allows to avoid the\nperformance degradation whenever the throughput measurement is not\nneeded, for example in production environment.\n\nNo device synchronize() is invoked when \"synchronized\" is set to False\n(default is True). This allows to produce approximate throughput\nmeasurements with minimal performance penalty.\n\n---------\n\nCo-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>\nCo-authored-by: Olatunji Ruwase ","shortMessageHtmlLink":"Add throughput timer configuration (#5363)"}},{"before":null,"after":"f4efef21b8370f5963f312a52968045b2f5e00b2","ref":"refs/heads/gh-readonly-queue/master/pr-5546-263bfe2892c1ee6285076214bb5e3898c35e78f3","pushedAt":"2024-05-22T20:23:09.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"[INF] DSAttention allow input_mask to have false as value (#5546)\n\nCo-authored-by: Logan Adams <114770087+loadams@users.noreply.github.com>","shortMessageHtmlLink":"[INF] DSAttention allow input_mask to have false as value (#5546)"}},{"before":"cec78ee1c9a14602775b4de80631ee2061e83b67","after":"430107751faf4f6a26a0ae8c9abe1f19b20c9c69","ref":"refs/heads/adk9/phi3-inference","pushedAt":"2024-05-22T17:15:05.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"loadams","name":"Logan Adams","path":"/loadams","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/114770087?s=80&v=4"},"commit":{"message":"Merge branch 'master' into adk9/phi3-inference","shortMessageHtmlLink":"Merge branch 'master' into adk9/phi3-inference"}},{"before":"94e1646246dad883c23901bfa28d97a4b3539392","after":null,"ref":"refs/heads/loadams/switch-hf-home","pushedAt":"2024-05-22T17:14:51.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"loadams","name":"Logan Adams","path":"/loadams","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/114770087?s=80&v=4"}},{"before":"263bfe2892c1ee6285076214bb5e3898c35e78f3","after":null,"ref":"refs/heads/gh-readonly-queue/master/pr-4816-29903925cf281e41742b6e5baba2daf6b83d2b76","pushedAt":"2024-05-22T17:13:30.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"}},{"before":"29903925cf281e41742b6e5baba2daf6b83d2b76","after":"263bfe2892c1ee6285076214bb5e3898c35e78f3","ref":"refs/heads/master","pushedAt":"2024-05-22T17:13:29.000Z","pushType":"merge_queue_merge","commitsCount":1,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Update to HF_HOME from TRANSFORMERS_CACHE (#4816)\n\nAddresses the following warning:\n\n```\n/tmp/actions-runner/_work/DeepSpeed/DeepSpeed/unit-test-venv/lib/python3.8/site-packages/transformers/utils/hub.py:123: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.\n```\n\nand the code on the transformers side is\n[here](https://github.com/huggingface/transformers/blob/1a585c1222a56bcaecc070966d558d4a9d862e83/src/transformers/utils/hub.py#L86C1-L96C81).","shortMessageHtmlLink":"Update to HF_HOME from TRANSFORMERS_CACHE (#4816)"}},{"before":null,"after":"263bfe2892c1ee6285076214bb5e3898c35e78f3","ref":"refs/heads/gh-readonly-queue/master/pr-4816-29903925cf281e41742b6e5baba2daf6b83d2b76","pushedAt":"2024-05-22T16:09:06.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"github-merge-queue[bot]","name":null,"path":"/apps/github-merge-queue","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9919?s=80&v=4"},"commit":{"message":"Update to HF_HOME from TRANSFORMERS_CACHE (#4816)\n\nAddresses the following warning:\n\n```\n/tmp/actions-runner/_work/DeepSpeed/DeepSpeed/unit-test-venv/lib/python3.8/site-packages/transformers/utils/hub.py:123: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.\n```\n\nand the code on the transformers side is\n[here](https://github.com/huggingface/transformers/blob/1a585c1222a56bcaecc070966d558d4a9d862e83/src/transformers/utils/hub.py#L86C1-L96C81).","shortMessageHtmlLink":"Update to HF_HOME from TRANSFORMERS_CACHE (#4816)"}},{"before":"e79d4b4dfa921b584795351080b6bb841e3124a7","after":"cec78ee1c9a14602775b4de80631ee2061e83b67","ref":"refs/heads/adk9/phi3-inference","pushedAt":"2024-05-22T01:42:42.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"adk9","name":"Abhishek Kulkarni","path":"/adk9","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/11399?s=80&v=4"},"commit":{"message":"Fix formatting","shortMessageHtmlLink":"Fix formatting"}},{"before":"3a678e80d71047124f801a360324e5cdb9e19920","after":"e79d4b4dfa921b584795351080b6bb841e3124a7","ref":"refs/heads/adk9/phi3-inference","pushedAt":"2024-05-21T23:50:57.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"adk9","name":"Abhishek Kulkarni","path":"/adk9","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/11399?s=80&v=4"},"commit":{"message":"Rename Phi to Phi3","shortMessageHtmlLink":"Rename Phi to Phi3"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAEU3bHOgA","startCursor":null,"endCursor":null}},"title":"Activity ยท microsoft/DeepSpeed"}