下面展示一个例子：

{
    "version": "0.2.0",
    "configurations": [

        {
            "name": "Python: Debug DeepSpeed",
            "type": "python",
            "request": "launch",
            "program": "/home/upa1/.conda/envs/dsmoe/bin/deepspeed",
            "justMyCode": true,
            "console": "integratedTerminal",
            "args": [
                "--num_nodes=1",
                "--num_gpus=1",
                "/data01/cym/Project/MoE/Megatron-DeepSpeed/Megatron-DeepSpeed/tools/generate_samples_gpt.py",
                "--tensor-model-parallel-size", "1",
                "--num-layers", "24",
                "--hidden-size", "1024",
                "--num-attention-heads", "16",
                "--max-position-embeddings", "1024",
                "--tokenizer-type", "GPT2BPETokenizer",
                "--fp16",
                "--num-experts", "1",
                "--mlp-type", "standard",
                "--micro-batch-size", "1",
                "--seq-length", "1024",
                "--out-seq-length", "1024",
                "--temperature", "1.0",
                "--vocab-file", "/data01/cym/Project/MoE/Megatron-DeepSpeed/Megatron-DeepSpeed/dataset/gpt2-vocab.json",
                "--merge-file", "/data01/cym/Project/MoE/Megatron-DeepSpeed/Megatron-DeepSpeed/dataset/gpt2-merges.txt",
                "--genfile", "unconditional_samples.json",
                "--top_p", "0.9",
                "--log-interval", "1",
                "--num-samples", "0",
                "--load", "/data01/cym/Project/MoE/Megatron-DeepSpeed/Megatron-DeepSpeed/dataset/checkpoints/gpt2_345m",
            ]
        }
        
    ]
}

• 其中program需要改成自己deepspeed所在位置，在运行环境中查看方式：which deeespeed
• args参数改成自己想要的就行。