projects

sd-llm-benchmark

No description available.

โญ Canonical โ€” this is the single source of truth for this project.

Overview

Field Value
Name sd-llm-benchmark
Tier ๐Ÿ—„๏ธ Archive
Description No description available.
Visibility ๐Ÿ”’ Private
Fork No
Last activity ย 
Status โœ… Archived
Original source Github
Original owner guneysus

๐Ÿ’ก Core Idea

A benchmarking tool for evaluating Large Language Model (LLM) performance on structured data tasks, including GPU-accelerated inference.

๐Ÿ” Code Snippet

"""
GPU Benchmarking Tool

This script benchmarks GPU performance for both:
1. Stable Diffusion image generation at specified resolutions and step counts
2. LLM inference speed in tokens per second

Usage:
    python gpu_benchmark.py --gpu [gpu_name] [other options]
"""

import argparse
import json
import os
import time
import platform
import torch
import numpy as np
from datetime import datetime
from tqdm import tqdm

# For Stable Diffusion
try:
    from diffusers import StableDiffusionPipeline
except ImportError:

Archive Origins


โ† ramblings _edit โ†’

Generated by scripts/generate.py ยท Back to index