PinnedInTDS ArchivebyMichaël Benesty·Nov 5, 2021Hugging Face Transformer Inference Under 1 Millisecond LatencyGo to production with Microsoft and Nvidia open source toolingA response icon17A response icon17
InTDS ArchivebyMichaël Benesty·May 20, 2020Divide Hugging Face Transformers training time by 2 or more…with dynamic padding and uniform length batchingA response icon2A response icon2
InTDS ArchivebyMichaël Benesty·Dec 10, 2019NER algo benchmark: spaCy, Flair, m-BERT and camemBERT on anonymizing French commercial legal casesDoes (model) size matters?A response icon8A response icon8
InTDS ArchivebyMichaël Benesty·Sep 26, 2019Why we switched from Spacy to Flair to anonymize French legal cases… and why you should always review your optionsA response icon4A response icon4