vllm.v1.worker ¶

Modules:

• block_table –
• cpu_model_runner –
• dp_utils –
• ec_connector_model_runner_mixin –

  Define EC connector functionality mixin for model runners.

• encoder_cudagraph –

  CUDA graph manager for vision encoder budget-batch execution.

• encoder_cudagraph_defs –

  Data transfer objects for encoder CUDA graph management.

• gpu –
• gpu_input_batch –
• gpu_model_runner –
• gpu_ubatch_wrapper –
• gpu_worker –

  A GPU worker class.

• kv_connector_model_runner_mixin –

  Define KV connector functionality mixin for model runners.

• lora_model_runner_mixin –

  Define LoRA functionality mixin for model runners.

• mamba_utils –
• tpu_input_batch –
• ubatch_utils –
• ubatching –
• utils –
• worker_base –
• workspace –
• xpu_model_runner –
• xpu_worker –