Excited to share the results from my latest experiment, which measures the inference performance of 20 open-source LLMs (from @AIatMeta, @MSFTResearch, @GoogleDeepMind, @MistralAI, @BigCodeProject) using @huggingface pipelines and @vllm_project (@woosuk_k) arxiv.org/pdf/2408.01050
Researcher, Software Engineer and University Teacher at @la_UPC

