Computational cost.

数据链接：

官方服务：

资源简介：

Number of trainable parameters for each model used for the downstream tasks. (XLSX)

创建时间：

2023-03-06

信息容量（Information Capacity）是一种基于文本压缩性能与计算复杂度相关性的大型语言模型（LLM）效率评估方法。该方法通过比较不同大小的模型在一系列任务中的表现，来预测模型性能。它还考虑了标记器效率对模型评估的影响。五个异构数据集被用于评估LLM的压缩效率。

Hugging Face2025-11-20 更新90

Network comparison w.r.t. parameter count, floating-point operations per second (FLOPs), and inference time.

Figshare2025-05-29 更新20

Comparison of the black box model and two-stage model in terms of efficiency.

NIAID Data Ecosystem10

Comparison of Parameters (M) and Flops (G) performance of different models on datasets.

Figshare2025-04-03 更新20

Performance comparison of different models in terms of model parameters, hardware resource usage, inference latency, and training time.

NIAID Data Ecosystem10