[Summary] 1. On-demand Container Loading in AWS Lambda introduces how to reduce scalability and cold start latency by introducing Block-level loading, Deduplication, and tiered cache. 2. Block-level loading makes an image into a single file system through a flattening process, then divides it into chunks and uploads it to the

[Summary] 1. BM25 is widely used for ranking method for IR. But since 2016, ranking methods using Neural Networks have started to appear 2. Neural IR generally showed high MRR, but had expensive computation. Especially BERT, which showed excellent MRR, was too slow to be applied to services 3. ColBERT