Back to evaluations
Public evaluation

hardware_allocation_task

Evaluates the AutoGen profiler's accuracy in hardware accelerator selection and MHS compliance generation.

Evaluation type
task based
Challenge
AutoGen MHS Workload Profiler & Accelerator Allocation Engine
Difficulty
Intermediate
Rigor
Unspecified

Evaluation overview

How the linked challenge is judged: tasks, benchmarks, and criteria count.

Tasks
1
Benchmarks
0
Criteria
0

Task templates

Inputs and expected outputs.

Task 1

hardware_allocation_task

Evaluates hardware recommendation for given model batch size, context length, and target SLA.

Input format

JSON containing model parameters (params_b, context_len, batch_size, SLA_ms)

Output format

JSON containing selected_accelerator, estimated_latency_ms, estimated_cost_usd, and MHS_profile