2.2.0

ascend-robot[pytorch][model]update qwen3_30b pp4ep4 to pp2ep8 in A2

e31e1bd6创建于 2025年12月2日历史提交

文件	最后提交记录	最后更新时间
ckpt_convert_qwen3_moe_235b_hf2mcore.sh	!3274 [pytorch][sh]add sh about ckpt of qwen3-235b Merge pull request !3274 from jwhk/master	8 个月前
ckpt_convert_qwen3_moe_235b_mcore2hf.sh	!3274 [pytorch][sh]add sh about ckpt of qwen3-235b Merge pull request !3274 from jwhk/master	8 个月前
ckpt_convert_qwen3_moe_hf2mcore.sh	!2768 add qwen3 moe dir& rename scripts Merge pull request !2768 from sunjunjie/master	11 个月前
ckpt_convert_qwen3_moe_mcore2hf.sh	!2828 [pytorch][sh]Add CUDA_DEVICE_MAX_CONNECTIONS Merge pull request !2828 from yanzhixiao/add-environment-variable	11 个月前
data_convert_qwen3_moe_instruction.sh	[pytorch][bugfix]in variable_seq_lengths mode, set --log-throughput to false Co-authored-by: sunjunjie1587<sunjunjie8@huawei.com>	7 个月前
data_convert_qwen3_moe_instruction_pack.sh	[pytorch][bugfix]in variable_seq_lengths mode, set --log-throughput to false Co-authored-by: sunjunjie1587<sunjunjie8@huawei.com>	7 个月前
data_convert_qwen3_moe_pairwise.sh	[pytorch][sh] update dpo in 2.2.0 Co-authored-by: yanzhixiao<yanzhixiao@h-partners.com> # message auto-generated for no-merge-commit merge: !3479 merge update-dpo-2.2.0 into 2.2.0 [pytorch][sh] update dpo in 2.2.0 Created-by: yanzhixiao23 Commit-by: yanzhixiao Merged-by: ascend-robot Description: update dpo in 2.2.0 See merge request: Ascend/MindSpeed-LLM!3479	7 个月前
data_convert_qwen3_moe_pretrain.sh	!2768 add qwen3 moe dir& rename scripts Merge pull request !2768 from sunjunjie/master	11 个月前
dpo_qwen3_30b_a3b_4K_A3_ptd.sh	[pytorch][sh] update dpo in 2.2.0 Co-authored-by: yanzhixiao<yanzhixiao@h-partners.com> # message auto-generated for no-merge-commit merge: !3479 merge update-dpo-2.2.0 into 2.2.0 [pytorch][sh] update dpo in 2.2.0 Created-by: yanzhixiao23 Commit-by: yanzhixiao Merged-by: ascend-robot Description: update dpo in 2.2.0 See merge request: Ascend/MindSpeed-LLM!3479	7 个月前
evaluate_qwen3_235b_a22b_ptd.sh	[pytorch][sh]fix qwen3 scripts error Co-authored-by: sunjunjie1587<sunjunjie8@huawei.com>	7 个月前
evaluate_qwen3_30b_a3b_ptd.sh	[pytorch][sh]fix qwen3 scripts error Co-authored-by: sunjunjie1587<sunjunjie8@huawei.com>	7 个月前
generate_qwen3_235b_a22b_ptd.sh	!3376 [pytorch][model] fix recompute_valid_actual_seq_len Merge pull request !3376 from jzh/master_recomactualseqlen	8 个月前
generate_qwen3_30b_a3b_ptd.sh	[pytorch][sh]fix qwen3 scripts error Co-authored-by: sunjunjie1587<sunjunjie8@huawei.com>	7 个月前
pretrain_qwen3_235b_a22b_4k_A3_ptd.sh	!3145 [pytroch][bugfix] add moe args in qwen3 Merge pull request !3145 from mhh001/master_0815	9 个月前
pretrain_qwen3_30b_a3b_4K_A3_fsdp2.sh	!3369 [pytorch][optimize]fix profile step setting and qwen3 scripts Merge pull request !3369 from 丁子叉/profile	8 个月前
pretrain_qwen3_30b_a3b_4K_ptd.sh	[pytorch][model]update qwen3_30b pp4ep4 to pp2ep8 in A2 Co-authored-by: guozhihua2<guozhihua2@huawei.com> # message auto-generated for no-merge-commit merge: !3832 merge update_qwen3_30b_sh_2.2.0 into 2.2.0 [pytorch][model]update qwen3_30b pp4ep4 to pp2ep8 in A2 Created-by: guozhihua2 Commit-by: guozhihua2 Merged-by: ascend-robot Description: 1. 更新qwen3_30b的切分pp4ep4到pp2ep8 See merge request: Ascend/MindSpeed-LLM!3832	5 个月前
tune_qwen3_235b_a22b_256K_full_pack_A3_ptd.sh	!3369 [pytorch][optimize]fix profile step setting and qwen3 scripts Merge pull request !3369 from 丁子叉/profile	8 个月前
tune_qwen3_235b_a22b_4K_full_pack_A3_ptd.sh	!3369 [pytorch][optimize]fix profile step setting and qwen3 scripts Merge pull request !3369 from 丁子叉/profile	8 个月前
tune_qwen3_235b_a22b_64K_full_pack_A3_ptd.sh	!3369 [pytorch][optimize]fix profile step setting and qwen3 scripts Merge pull request !3369 from 丁子叉/profile	8 个月前
tune_qwen3_30b_a3b_4K_full_ptd.sh	[pytorch][sh]fix qwen3 scripts error Co-authored-by: sunjunjie1587<sunjunjie8@huawei.com>	7 个月前
tune_qwen3_30b_a3b_4K_lora_ptd.sh	[pytorch][sh]fix qwen3 scripts error Co-authored-by: sunjunjie1587<sunjunjie8@huawei.com>	7 个月前