CtrlK
BlogDocsLog inGet started
Tessl Logo

tessl/pypi-vllm

A high-throughput and memory-efficient inference and serving engine for LLMs

Overall
score

69%

Evaluation69%

1.33x

Agent success when using this tile

Overview
Eval results
Files

rubric.jsonevals/scenario-3/

{
  "context": "This criteria evaluates how well the engineer uses vLLM's memory management features to configure LLM instances with custom GPU memory utilization and CPU swap space settings.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "LLM class import",
      "description": "Correctly imports the LLM class from the vllm package",
      "max_score": 10
    },
    {
      "name": "LLM initialization",
      "description": "Creates LLM instances using the LLM constructor with the model_name parameter",
      "max_score": 15
    },
    {
      "name": "gpu_memory_utilization parameter",
      "description": "Uses the gpu_memory_utilization parameter in the LLM constructor to control GPU memory allocation (passing float values like 0.7, 0.8)",
      "max_score": 25
    },
    {
      "name": "swap_space parameter",
      "description": "Uses the swap_space parameter in the LLM constructor to configure CPU swap space (passing integer values representing GB)",
      "max_score": 25
    },
    {
      "name": "Combined configuration",
      "description": "Correctly combines both gpu_memory_utilization and swap_space parameters in a single LLM initialization",
      "max_score": 15
    },
    {
      "name": "Default configuration",
      "description": "Handles cases where no memory parameters are specified, allowing vLLM to use default settings",
      "max_score": 10
    }
  ]
}

Install with Tessl CLI

npx tessl i tessl/pypi-vllm

tile.json