infra-bench/infra-bench-v1
Coding
Professional
20 AI-infra engineering tasks on AMD MI35X (5 categories: model deployment, profiling, kernel implementation, kernel tuning, debug triage).
Run this task
CLI:
inspect eval inspect_harbor/infra_bench_v1 --model openai/gpt-5Python:
from inspect_ai import eval
from inspect_harbor import infra_bench_v1
eval(infra_bench_v1(), model="openai/gpt-5")Dataset information
| Harbor registry | infra-bench/infra-bench-v1 |
| Inspect task | infra_bench_v1 |
| Latest digest | sha256:ee4dab2ad2279aaf0bc52c2736083844e9555ef86236e8ee7867bd0ba7afde87 |
| Samples | 20 |
| Source | https://github.com/sammysun0711/InfraBench |
See Task Parameters for the parameter set shared across all Harbor tasks.