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 259664 次贡献
更少更多
- frostbyte_neo 推送了仓库 github-featured/nvidia-nemo--speech 的 maanug/add-cleanup-util 分支f9c1bd769f add mcore global state cleanup fn
- frostbyte_neo 推送了仓库 github-featured/nvidia-nemo--speech 的 longform_speechlm2_chunk 分支比较 3 提交 »7408b73bcc Apply isort and black reformatting4a8f11067f add longform chunk inference9cd7bfa145 Merge branch 'main' into salm-with-transcript-input-sharegpt
- frostbyte_neo 推送了仓库 github-featured/nvidia-nemo--speech 的 yash/llava_repro 分支e0f1d6b0e5 changes for llava repro
- frostbyte_neo 推送了仓库 github-featured/nvidia-nemo--speech 的 llmb-nemo-r2.5.0 分支比较 8 提交 »0406e44b83 Add gb300 to numa ctl pinning logic. (#15516)032e4de5b7 Enable Offline mode for Nemotron 15b and 340b tokenizer. (#15196)d08a142547 Remove NCCL_NET_GDR_LEVEL and NCCL_NET_GDR_C2C environment variables. (#15160)9957a50be3 Extend Slurm exit code override to additional LLM pretrain recipes (#15085)c02d425d18 Set segment for gb systems when nodes <= 18. (#15060)
- frostbyte_neo 推送了仓库 github-featured/nvidia-nemo--speech 的 yash/llava_next_fixes 分支7c80644a7c use_fast as true for tokenizer in lalva
- frostbyte_neo 推送了仓库 github-featured/nvidia-nemo--speech 的 yash/llava_next_data 分支比较 2 提交 »b832c972e0 added doc strings for LlavaNextTextSample LlavaNextTextRawBatchfa58eefde7 LLava next data module changes for energon and mock
- frostbyte_neo 推送了仓库 github-featured/nvidia-nemo--speech 的 yash/energon_datamodule_refactor 分支1a4eaae5cf refactored energon datamodule to remove image_processor and task encoder from init
