HyperAI超神经

Language Modelling On C4

评估指标

Perplexity
Steps
TPUv3 Hours

评测结果

各个模型在此基准测试上的表现结果

比较表格
模型名称PerplexityStepsTPUv3 Hours
primer-searching-for-efficient-transformers12.691M16.5K
llm-int8-8-bit-matrix-multiplication-for12.45--
primer-searching-for-efficient-transformers13.251M15.7K
n-grammer-augmenting-transformers-with-latent-114.79--
llm-int8-8-bit-matrix-multiplication-for14.43--
llm-int8-8-bit-matrix-multiplication-for15.91--
primer-searching-for-efficient-transformers12.351M17.3K
n-grammer-augmenting-transformers-with-latent-115.01--
llm-int8-8-bit-matrix-multiplication-for13.3--