11月12月1月2月3月4月5月6月7月8月9月10月
周一周三周五
一年内 256794 次贡献
更少更多
- frostbyte_neo 推送了仓库 github-featured/dmlc--dgl 的 stanyang_probmask 分支比较 2 提交 »6a024733f8 remove squeeze(1)1fc5f8a923 change based on #7462
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/write_checkpoints 分支比较 10 提交 »a13b2bd10b fix bug check null2a0f78d617 attempt to fix windows episode 25b96d91893 attempt to fix windows?f2ee356a06 resolve merge conflict4ff0412751 Merge pull request #450 from rosslwheeler/dataloader_win_fixes
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/wikitext103 分支比较 10 提交 »3048a3a780 oops have to use labels=b4278f094a v1 of what was supposed to be a repro of wikitext ppl, but isn't, what6798dd4b91 Merge branch 'rosslwheeler-Makefile-detect-gpu'5344cc30b8 Merge branch 'Makefile-detect-gpu' of https://github.com/rosslwheeler/llm.c into Makefile-detect-gpu0c4908d815 Adding CI check to disable auto-detect
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/streams 分支比较 10 提交 »ee6b3c9325 take out Async copies and memsets6d14c29df5 streams, everything right now is on main stream, almostd5f7458e03 Merge branch 'stream-2' of https://github.com/ngc92/llm.c into ngc92-stream-29af7628063 moved remaining memcpy/memset in the training loop to main_stream8940dcdcb1 Merge branch 'stream-2' of https://github.com/ngc92/llm.c into ngc92-stream-2
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/weight_decay_2d_only 分支比较 10 提交 »9f08882051 add weight decay, but only for 2D tensors, as done in GPT series and in general too. this forces us to break up our adamw kernel again into one call per tensor, so there is a small throughput hit, of about 0.5% for me. but we have to break up this kernel in near future anywaye60c4843a3 Merge branch 'ngc92-cpu-matmul'3221e4b2d2 small cosmetic changes6e4296fb9a Merge branch 'cpu-matmul' of https://github.com/ngc92/llm.c into ngc92-cpu-matmulb8674cedf8 Merge branch 'ademeure-deterministic_encoder_v2'
- frostbyte_neo 推送了仓库 github-featured/dmlc--dgl 的 stanyang 分支比较 2 提交 »b4df206c64 fix issue #77669e6602f698 merge origin
- frostbyte_neo 推送了仓库 github-featured/dmlc--dgl 的 revert-7155-argo_tutorial 分支c6df1c4f76 Revert "[Doc] Created a tutorial for ARGO in the CPU section (#7155)"
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/rolling_checkpoints 分支比较 10 提交 »11e2fd1cf2 separate out write_checkpoint and delete_checkpoint8952f0235b rolling checkpoints - checkpoints are either MINOR or MAJOR and minor checkpoints get deleted with a rolling window. This is an optimization that will allow us to save state more often, but preserve disk space overall. And if we ever see loss spikes and such, it is easier to reset the state to an earlier checkpoint that is more recent76b133d3f4 Merge pull request #631 from gordicaleksa/minor_refactor269b50adcf5 Merge pull request #632 from gordicaleksa/multi_node_my4af6a6ab3d Remove /tmp as default init value for fs_path
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/sharded_data 分支比较 10 提交 »099d30f814 add a super small crappy glob for windows that only matches a single unique file. this will make CI happy but we can't train on sharded data on windows until this is improved05be4f6d82 readme changesedb0df967a continued changes for sharded dataloader31310282e1 extend dataloader to be sharded967420d1d1 fix print format warning for size_t vs int
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/resume 分支比较 10 提交 »24a8c03cb6 Merge branch 'master' into feature/resume69d0583ac8 conditionally include dirent on not windows3b4d8bece6 Merge pull request #474 from rosslwheeler/Windows-dirent-fixbb80f93618 Minor cleanup758acdffbf Adding dirent.h support to unistd.h
- frostbyte_neo 推送了仓库 github-featured/karpathy--llm.c 的 feature/restore_from_master 分支比较 10 提交 »51dd102328 remove confusing comment2b827f1659 bring back state allocation into build_from_checkpointa794bcb395 small fixesb2ae84745d Merge branch 'restore_from_master_weights' of https://github.com/ademeure/llm.c into ademeure-restore_from_master_weightsb4623bc5e7 Merge pull request #712 from indianspeedster/dev-cuda
