[bug][AutoDeploy]: Sharding path fails with IndexError: list index out of range

System Info

• CPU architecture: x86_64
• GPU properties: not captured from CI artifact
• Libraries
  • TensorRT-LLM branch or tag: main (pipeline context)
  • Container used: AutoDeploy CI container
• OS: CI environment

Who can help?

No response

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

Affected registry entries currently disabled in examples/auto_deploy/model_registry/models.yaml:

• CohereLabs/c4ai-command-a-03-2025
• CohereLabs/command-a-reasoning-08-2025
• CohereLabs/command-a-translate-08-2025
• Qwen/Qwen2-VL-72B-Instruct-GPTQ-Int4
• HuggingFaceTB/SmolLM2-1.7B-Instruct
• google/codegemma-7b-it
• google/gemma-1.1-7b-it

I fetched the per-test stdout logs for all seven models from terminal pipeline 52785950. All seven fail in the AutoDeploy sharding path with the same upper traceback through:

• tensorrt_llm/_torch/auto_deploy/transform/library/sharding.py:1272
• tensorrt_llm/_torch/auto_deploy/transform/library/sharding.py:3204

Six models then bottom out with:

IndexError: list index out of range

in tensorrt_llm/_torch/auto_deploy/utils/node_utils.py:1436 while computing weight shape inside get_all_layer_subgraphs.

The remaining model (Qwen/Qwen2-VL-72B-Instruct-GPTQ-Int4) still fails in the same sharding path but hits a slightly different index access in node_utils.py:1032, also with:

IndexError: list index out of range

This looks like one shared sharding-family issue rather than seven unrelated model-specific failures.

Expected behavior

The AutoDeploy sharding analysis should handle these models without hitting IndexError while discovering layer subgraphs or weight shapes.

actual behavior

The AutoDeploy sharding path fails with IndexError: list index out of range across multiple models before successful execution.

additional notes

NA

Before submitting a new issue...

• Make sure you already searched for relevant issues, and checked the documentation and examples for answers to frequently asked questions.

Start by reproducing the failures for the affected registry entries and inspect examples/auto_deploy/model_registry/models.yaml. Trace the sharding path at tensorrt_llm/_torch/auto_deploy/transform/library/sharding.py:1272 and :3204, then inspect node_utils.py:1436 and :1032. Done means the listed models complete sharding analysis without IndexError.