2.0.0

文件	最后提交记录	最后更新时间
ckpt_convert_deepseek2_hf2mcore.sh	!2076 新增DeepSeek-V2.5 8k/32k预训练 Merge pull request !2076 from shengjy/master	1 年前
ckpt_convert_deepseek2_mcore2hf.sh	!1998 rename: repo package name from modellink to mindspeed_llm Merge pull request !1998 from MeiFei/master-package-rename	1 年前
data_convert_deepseek2_pretrain.sh	!1710 补充公网地址声明 Merge pull request !1710 from guoxinjie/master	1 年前
evaluate_deepseek2_236b_ptd.sh	!2352 fix some GPUS_PER_NODE to NPUS_PER_NODE Merge pull request !2352 from jzh/master-v3example	1 年前
generate_deepseek2_236b_ptd.sh	!2352 fix some GPUS_PER_NODE to NPUS_PER_NODE Merge pull request !2352 from jzh/master-v3example	1 年前
pretrain_deepseek2_100b_8k_A3_ptd.sh	!2581 [双合]告警dsk非全参模型 Merge pull request !2581 from jzh/2.0.0	1 年前
pretrain_deepseek2_100b_8k_ptd.sh	!2581 [双合]告警dsk非全参模型 Merge pull request !2581 from jzh/2.0.0	1 年前
pretrain_deepseek2_236b_8k_A3_ptd.sh	!2499 修复部分脚本缺少ckpt_load,ckpt_save的问题 Merge pull request !2499 from jzh/2.0.0sh	1 年前
pretrain_deepseek2_236b_8k_ptd.sh	!2499 修复部分脚本缺少ckpt_load,ckpt_save的问题 Merge pull request !2499 from jzh/2.0.0sh	1 年前
pretrain_deepseek2_60b_128k_A3_ptd.sh	!2581 [双合]告警dsk非全参模型 Merge pull request !2581 from jzh/2.0.0	1 年前
pretrain_deepseek2_60b_8k_ptd.sh	!2581 [双合]告警dsk非全参模型 Merge pull request !2581 from jzh/2.0.0	1 年前
pretrain_deepseek2_ptd_8p.sh	!2581 [双合]告警dsk非全参模型 Merge pull request !2581 from jzh/2.0.0	1 年前