### 0
[multi_node num_nodes=None] PYTHONPATH=<REPO_ROOT>:/root/Megatron-LM:/frozen/pythonpath torchrun
  --nproc-per-node 8
  --master-addr {{master_addr}}
  --master-port 23456
  --nnodes={{nnodes}}
  --node-rank {{node_rank}} <REPO_ROOT>/tools/convert_hf_to_torch_dist.py
  --spec miles_plugins.models.glm5.glm5 get_glm5_spec
  --disable-bias-linear
  --num-layers 61
  --hidden-size 7168
  --ffn-hidden-size 18432
  --num-attention-heads 128
  --kv-channels 128
  --normalization RMSNorm
  --position-embedding-type rope
  --norm-epsilon 1e-6
  --swiglu
  --untie-embeddings-and-output-weights
  --vocab-size 129280
  --multi-latent-attention
  --q-lora-rank 1536
  --kv-lora-rank 512
  --qk-head-dim 128
  --qk-pos-emb-head-dim 64
  --v-head-dim 128
  --qk-layernorm
  --rotary-scaling-factor 40
  --rotary-base 10000
  --mscale 1.0
  --mscale-all-dim 1.0
  --attention-softmax-in-fp32
  --no-rope-fusion
  --num-experts 256
  --moe-layer-freq '[0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]'
  --moe-ffn-hidden-size 2048
  --moe-router-topk 8
  --moe-shared-expert-intermediate-size 2048
  --moe-router-pre-softmax
  --moe-router-score-function sigmoid
  --moe-router-enable-expert-bias
  --moe-router-load-balancing-type seq_aux_loss
  --moe-token-dispatcher-type alltoall
  --moe-aux-loss-coeff 0
  --moe-router-bias-update-rate 0
  --moe-router-group-topk 4
  --moe-router-num-groups 8
  --moe-grouped-gemm
  --moe-router-topk-scaling-factor 2.5
  --moe-router-dtype fp32
  --moe-permute-fusion
  --hf-checkpoint /root/models/DeepSeek-V3.2-bf16
  --save /root/models/DeepSeek-V3.2_torch_dist
  --tensor-model-parallel-size 4
  --expert-model-parallel-size 16
  --pipeline-model-parallel-size 6
  --decoder-last-pipeline-num-layers 13
  --expert-tensor-parallel-size 1 
