12月1月2月3月4月5月6月7月8月9月10月11月
周一周三周五
一年内 260010 次贡献
更少更多
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 olruwase/zero_optional_reduce_scatter 分支980fdeda17 Merge branch 'master' into olruwase/zero_optional_reduce_scatter
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 gh-readonly-queue/master/pr-7908-5f7b687018bd1e0340c661859820fd97aa80a616 分支169c828af6 Add Feature Universal Checkpoint for AutoTP (#7908)
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 olruwase/zero_offload_v3 分支133730d82b Merge branch 'staging-zero-dual-v3' of github.com:microsoft/DeepSpeed into olruwase/zero_offload_v3
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 olruwase/zero_offload_fix_corner_case 分支比较 3 提交 »d11643375f Increate bucket size scalerdbcebb83dd Format fixes80b99c6a3a Add ZeRO-Offload model tests
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 gh-readonly-queue/master/pr-4163-5e16eb2c939707d0d0062a458d77998fccb3afad 分支817eef0594 Fix nv-nightly workflow (#4163)
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 olruwase/zero_offload_e2e 分支比较 2 提交 »65ba77bbe8 Make ZeRO support DeepSpeedCPUAdambd429188cf Merge branch 'staging-zero-dual-v2' into olruwase/zero_offload_e2e
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 olruwase/zero_inference_torch_version 分支f558c9b40d Fix min torch version
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 gh-readonly-queue/master/pr-3928-82115d9059ce8271229c8f63153a02f2d323cfc1 分支比较 2 提交 »75facef435 remove the call to param.ds_tensor from print (#3928)82115d9059 Reduce Unit Test Times (Part 3) (#3850)
- frostbyte_neo 推送了仓库 github-featured/deepspeedai--deepspeed 的 gh-readonly-queue/master/pr-3893-cc71eec8c85c4437d8139e53372da7f22224fed5 分支e725d66912 avoid init for deepspeed backend first (#3893)
