[Bug]: OOM in meta-llama_Llama-3.3-70B-Instruct accuracy test

System Info

See nvbug: 5981841

Who can help?

@marinayanov PTAL if you have the capacity

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

pytest ./tests/integration/defs/accuracy/test_llm_api_autodeploy.py::TestModelRegistryAccuracy::test_autodeploy_from_registry[meta-llama_Llama-3.3-70B-Instruct-False]

Expected behavior

Test passes

actual behavior

[03/14/2026-18:47:16] [TRT-LLM] [RANK 3] [E] Failed to initialize executor on rank 3: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 3 has a total capacity of 79.10 GiB of which 477.88 MiB is free. Including non-PyTorch memory, this process has 78.62 GiB memory in use. Of the allocated memory 76.70 GiB is allocated by PyTorch, and 944.31 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

additional notes

70B BF16 model - running on 80GB is not enough. Need to increase minimum memory for this test.

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 running pytest ./tests/integration/defs/accuracy/test_llm_api_autodeploy.py::TestModelRegistryAccuracy::test_autodeploy_from_registry[meta-llama_Llama-3.3-70B-Instruct-False] and inspect the referenced test's model and resource configuration. Done means the 70B BF16 accuracy test requires sufficient GPU memory and passes on supported hardware without the reported 80GB out-of-memory failure.