{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":783833344,"defaultBranch":"master","name":"llm.c","ownerLogin":"karpathy","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2024-04-08T16:58:11.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/241138?v=4","public":true,"private":false,"isOrgOwned":false},"refInfo":{"name":"","listCacheKey":"v0:1715886565.0","currentOid":""},"activityList":{"items":[{"before":"130712215df6d67049fbc4179cc217f5a1d4e69f","after":"9e04deabe3e519ea3420be7a6fcb6067ee89334b","ref":"refs/heads/master","pushedAt":"2024-05-16T20:12:55.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #416 from azret/feature/cpu_dev\n\nSetting up dev/cpu area with the first matmul_forward.c","shortMessageHtmlLink":"Merge pull request #416 from azret/feature/cpu_dev"}},{"before":"68eea0bd5f9442624ededa8b156f46cae21df889","after":"130712215df6d67049fbc4179cc217f5a1d4e69f","ref":"refs/heads/master","pushedAt":"2024-05-16T20:04:07.000Z","pushType":"push","commitsCount":4,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge branch 'backward-bias' of https://github.com/ngc92/llm.c","shortMessageHtmlLink":"Merge branch 'backward-bias' of https://github.com/ngc92/llm.c"}},{"before":"bd7dc7af9688c17d134f48a5269c7c5d669bce14","after":"68eea0bd5f9442624ededa8b156f46cae21df889","ref":"refs/heads/master","pushedAt":"2024-05-16T19:49:03.000Z","pushType":"push","commitsCount":5,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Fix ZeRO-1, we were communicating way too much","shortMessageHtmlLink":"Fix ZeRO-1, we were communicating way too much"}},{"before":"fbc5cba1f954a3e3d6c222724ab9ef5b43041047","after":"bd7dc7af9688c17d134f48a5269c7c5d669bce14","ref":"refs/heads/master","pushedAt":"2024-05-16T19:25:31.000Z","pushType":"pr_merge","commitsCount":4,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #422 from karpathy/feature/recompute\n\nfeature/recompute","shortMessageHtmlLink":"Merge pull request #422 from karpathy/feature/recompute"}},{"before":null,"after":"d7581fc5428b02d71b954ba9f9f073627c4d0a83","ref":"refs/heads/feature/recompute","pushedAt":"2024-05-16T19:09:25.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"make recompute be an int instead of bool, so we can strengthen it over time just like ZeRO stages, as we recompute more and more of the model in the future possibly. and make it default on because it is awesome","shortMessageHtmlLink":"make recompute be an int instead of bool, so we can strengthen it ove…"}},{"before":"0f48a28f12562f1614d9747dfd7f2ca0c83eb7b5","after":"fbc5cba1f954a3e3d6c222724ab9ef5b43041047","ref":"refs/heads/master","pushedAt":"2024-05-16T09:45:22.000Z","pushType":"pr_merge","commitsCount":3,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #415 from azret/feature/eot\n\ntrain_gpt.c: Removing the hardcoded GPT2_EOT","shortMessageHtmlLink":"Merge pull request #415 from azret/feature/eot"}},{"before":"2d43e5bc977bac537ae1f6aff7de978fba328409","after":"0f48a28f12562f1614d9747dfd7f2ca0c83eb7b5","ref":"refs/heads/master","pushedAt":"2024-05-15T21:53:16.000Z","pushType":"pr_merge","commitsCount":4,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #412 from karpathy/feature/grad_accum3\n\n[wip] gradient accumulation, another attempt","shortMessageHtmlLink":"Merge pull request #412 from karpathy/feature/grad_accum3"}},{"before":"92fc26eba4549057b40726d03e36d483e40542da","after":"8eb3a432d991bf23bed3fae9f92dd395015ea457","ref":"refs/heads/feature/grad_accum3","pushedAt":"2024-05-15T21:11:43.000Z","pushType":"push","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"revert all streams and synchronization events, we'll bring them back but only one at a time and very very carefully, because they cause really subtle bugs and issues","shortMessageHtmlLink":"revert all streams and synchronization events, we'll bring them back …"}},{"before":null,"after":"92fc26eba4549057b40726d03e36d483e40542da","ref":"refs/heads/feature/grad_accum3","pushedAt":"2024-05-14T21:32:53.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"the nuts and bolts of gradient accumulation again, merged to master, but there is a bug and it doesn't work, debugging...","shortMessageHtmlLink":"the nuts and bolts of gradient accumulation again, merged to master, …"}},{"before":"222d59fa2f6bd40b0ecba8d32e041cb032b585e2","after":"2d43e5bc977bac537ae1f6aff7de978fba328409","ref":"refs/heads/master","pushedAt":"2024-05-14T19:14:12.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"remove legacy comment","shortMessageHtmlLink":"remove legacy comment"}},{"before":"2346cdac931f544d63ce816f7e3f5479a917eef5","after":"222d59fa2f6bd40b0ecba8d32e041cb032b585e2","ref":"refs/heads/master","pushedAt":"2024-05-14T19:13:09.000Z","pushType":"pr_merge","commitsCount":3,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #408 from ngc92/layernorm-bw-dev-cuda\n\nLayernorm backward updates","shortMessageHtmlLink":"Merge pull request #408 from ngc92/layernorm-bw-dev-cuda"}},{"before":"750c5fd7e8d215f09af33a3736374ca97cc4ac27","after":"2346cdac931f544d63ce816f7e3f5479a917eef5","ref":"refs/heads/master","pushedAt":"2024-05-13T20:19:16.000Z","pushType":"push","commitsCount":10,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge branch 'ngc92-backward-bias'","shortMessageHtmlLink":"Merge branch 'ngc92-backward-bias'"}},{"before":"c1814d5d2fc346b1a72ac3cd4448a743c675572c","after":"750c5fd7e8d215f09af33a3736374ca97cc4ac27","ref":"refs/heads/master","pushedAt":"2024-05-13T20:08:53.000Z","pushType":"pr_merge","commitsCount":17,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #309 from ahrefs/zero-stage1\n\nZero Redundancy Optimizer - Stage1","shortMessageHtmlLink":"Merge pull request #309 from ahrefs/zero-stage1"}},{"before":"a2bffb5a8a3aec4127063c787eb9add66ac8f979","after":"c1814d5d2fc346b1a72ac3cd4448a743c675572c","ref":"refs/heads/master","pushedAt":"2024-05-12T16:32:14.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #401 from rosslwheeler/ci-add-win\n\nAdd Windows to CI","shortMessageHtmlLink":"Merge pull request #401 from rosslwheeler/ci-add-win"}},{"before":"1cfe899121d126a90d8df4b241a81b22573e3aa7","after":"a2bffb5a8a3aec4127063c787eb9add66ac8f979","ref":"refs/heads/master","pushedAt":"2024-05-12T16:29:20.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #403 from nietras/nietras-llm-cs\n\nAdd Llm.cs by nietras to README","shortMessageHtmlLink":"Merge pull request #403 from nietras/nietras-llm-cs"}},{"before":"5dbda7e1156d0f8fa68a5f55a9c898eff8971222","after":"1cfe899121d126a90d8df4b241a81b22573e3aa7","ref":"refs/heads/master","pushedAt":"2024-05-11T17:21:23.000Z","pushType":"push","commitsCount":5,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge branch 'master' of github.com:karpathy/llm.c","shortMessageHtmlLink":"Merge branch 'master' of github.com:karpathy/llm.c"}},{"before":"923b2b0a8fd5d39a949ff99fdb050d2fdba23a8e","after":"5dbda7e1156d0f8fa68a5f55a9c898eff8971222","ref":"refs/heads/master","pushedAt":"2024-05-11T17:03:40.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #397 from lancerts/comment-fix\n\nUpdate the -lcublas -lcublasLt flag in the comment","shortMessageHtmlLink":"Merge pull request #397 from lancerts/comment-fix"}},{"before":"fd007d6508ac22cdf827291e8fcd3778f0c8dd30","after":"923b2b0a8fd5d39a949ff99fdb050d2fdba23a8e","ref":"refs/heads/master","pushedAt":"2024-05-10T10:13:53.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #398 from zwets/fix-cudnn-frontend-check\n\nFix detection of cudnn-frontend in '.' on Linux","shortMessageHtmlLink":"Merge pull request #398 from zwets/fix-cudnn-frontend-check"}},{"before":"83ad9ac87fad31a9e02fad823f099edf8015f8e1","after":"fd007d6508ac22cdf827291e8fcd3778f0c8dd30","ref":"refs/heads/master","pushedAt":"2024-05-09T23:18:25.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #395 from yijunyu/patch-1\n\nUpdate README.md","shortMessageHtmlLink":"Merge pull request #395 from yijunyu/patch-1"}},{"before":"077b649ee110644e56f27f19d715a7222103f68e","after":"83ad9ac87fad31a9e02fad823f099edf8015f8e1","ref":"refs/heads/master","pushedAt":"2024-05-09T23:17:35.000Z","pushType":"push","commitsCount":6,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Fused Residual+LayerNorm layers, ~1% faster","shortMessageHtmlLink":"Fused Residual+LayerNorm layers, ~1% faster"}},{"before":"b8eaafdf9b1ed95aa95456125360c080081243fc","after":"077b649ee110644e56f27f19d715a7222103f68e","ref":"refs/heads/master","pushedAt":"2024-05-09T18:30:48.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #389 from vedaanta/cudnn/install\n\nAdds cuDNN install instructions for a smaller but sufficient package","shortMessageHtmlLink":"Merge pull request #389 from vedaanta/cudnn/install"}},{"before":"68509c89a25e5c1e28299eed670cf1bc875473b9","after":"b8eaafdf9b1ed95aa95456125360c080081243fc","ref":"refs/heads/master","pushedAt":"2024-05-08T21:00:57.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #386 from Anerudhan/feature/rename_cudnn_attn\n\nRename cudnn_att.cu to cudnn_att.cpp to speed up compilation.","shortMessageHtmlLink":"Merge pull request #386 from Anerudhan/feature/rename_cudnn_attn"}},{"before":"26dbbc75c29b618d0038c92776424124e57d0598","after":"68509c89a25e5c1e28299eed670cf1bc875473b9","ref":"refs/heads/master","pushedAt":"2024-05-08T20:20:30.000Z","pushType":"push","commitsCount":3,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge branch 'pjj-fix-gpt2-xl'","shortMessageHtmlLink":"Merge branch 'pjj-fix-gpt2-xl'"}},{"before":"84a2aa0dd08039493a7226c505bad919e594b870","after":"26dbbc75c29b618d0038c92776424124e57d0598","ref":"refs/heads/master","pushedAt":"2024-05-08T20:18:41.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"unbreak the tests oops","shortMessageHtmlLink":"unbreak the tests oops"}},{"before":"1c5ef8cea3599b858baf8b5f8913c0e5878c30e7","after":"84a2aa0dd08039493a7226c505bad919e594b870","ref":"refs/heads/master","pushedAt":"2024-05-08T20:13:17.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"allow exporting all GPT-2 model sizes now, using --model flag in python script. also allow loading all the models from C, using the new flag -e, to point directly to the file to load. added some error handling for potentially common mistakes. note that loadin GPT-2 XL does not work yet and crashes, but I am imminently merging a fix as the next commit that resolves this","shortMessageHtmlLink":"allow exporting all GPT-2 model sizes now, using --model flag in pyth…"}},{"before":"2f6c54513105a30a9ecfd408e180b223e6a6ff91","after":"1c5ef8cea3599b858baf8b5f8913c0e5878c30e7","ref":"refs/heads/master","pushedAt":"2024-05-08T17:50:00.000Z","pushType":"pr_merge","commitsCount":3,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #382 from ahrefs/fix-multigpu-hang\n\nFix incorrect GPU assignment in multi gpu setup","shortMessageHtmlLink":"Merge pull request #382 from ahrefs/fix-multigpu-hang"}},{"before":"0141408ca544b7ffbaf7e5d38bd7e24a3b5a8481","after":"2f6c54513105a30a9ecfd408e180b223e6a6ff91","ref":"refs/heads/master","pushedAt":"2024-05-07T09:36:40.000Z","pushType":"push","commitsCount":6,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge branch 'ademeure-more_kernel_opt'","shortMessageHtmlLink":"Merge branch 'ademeure-more_kernel_opt'"}},{"before":"ce72242c46f87927331bf5e59979690f8b808635","after":"0141408ca544b7ffbaf7e5d38bd7e24a3b5a8481","ref":"refs/heads/master","pushedAt":"2024-05-07T00:50:11.000Z","pushType":"pr_merge","commitsCount":6,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #357 from rosslwheeler/cudnn_makefile_changes\n\ncuDNN makefile changes","shortMessageHtmlLink":"Merge pull request #357 from rosslwheeler/cudnn_makefile_changes"}},{"before":"3cc16f135412b0269cc6b369078fd50bdd4ebc89","after":"ce72242c46f87927331bf5e59979690f8b808635","ref":"refs/heads/master","pushedAt":"2024-05-07T00:47:47.000Z","pushType":"pr_merge","commitsCount":5,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"Merge pull request #376 from rosslwheeler/train_gp2cu-porting-fixes\n\nReplace non-standard C syntax with standard C syntax - fix win compilation","shortMessageHtmlLink":"Merge pull request #376 from rosslwheeler/train_gp2cu-porting-fixes"}},{"before":"ec9fd51d5c9c244c9a5dc8f879f4aa7eb64c9d19","after":"3cc16f135412b0269cc6b369078fd50bdd4ebc89","ref":"refs/heads/master","pushedAt":"2024-05-07T00:39:47.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"karpathy","name":"Andrej","path":"/karpathy","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/241138?s=80&v=4"},"commit":{"message":"fix logits bug","shortMessageHtmlLink":"fix logits bug"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAETAhuWwA","startCursor":null,"endCursor":null}},"title":"Activity · karpathy/llm.c"}