HPLT × NLPL Winter School · Day 2 · 11 February 2025 · Scroll to read
About these notes
Lecture material on building a large multilingual pre-training corpus: data quality, filtering, deduplication, language identification, tokenization, experiments, and evaluation.
- Multilingual data quality and filtering
- Deduplication and language identification
- FineWeb 2 construction and evaluation
-
Page 1 of 74 -
Page 2 of 74 -
Page 3 of 74 -
Page 4 of 74 -
Page 5 of 74 -
Page 6 of 74 -
Page 7 of 74 -
Page 8 of 74 -
Page 9 of 74 -
Page 10 of 74 -
Page 11 of 74 -
Page 12 of 74 -
Page 13 of 74 -
Page 14 of 74 -
Page 15 of 74 -
Page 16 of 74 -
Page 17 of 74 -
Page 18 of 74 -
Page 19 of 74 -
Page 20 of 74 -
Page 21 of 74 -
Page 22 of 74 -
Page 23 of 74 -
Page 24 of 74 -
Page 25 of 74 -
Page 26 of 74 -
Page 27 of 74 -
Page 28 of 74 -
Page 29 of 74 -
Page 30 of 74 -
Page 31 of 74 -
Page 32 of 74 -
Page 33 of 74 -
Page 34 of 74 -
Page 35 of 74 -
Page 36 of 74 -
Page 37 of 74 -
Page 38 of 74 -
Page 39 of 74 -
Page 40 of 74 -
Page 41 of 74 -
Page 42 of 74 -
Page 43 of 74 -
Page 44 of 74 -
Page 45 of 74 -
Page 46 of 74 -
Page 47 of 74 -
Page 48 of 74 -
Page 49 of 74 -
Page 50 of 74 -
Page 51 of 74 -
Page 52 of 74 -
Page 53 of 74 -
Page 54 of 74 -
Page 55 of 74 -
Page 56 of 74 -
Page 57 of 74 -
Page 58 of 74 -
Page 59 of 74 -
Page 60 of 74 -
Page 61 of 74 -
Page 62 of 74 -
Page 63 of 74 -
Page 64 of 74 -
Page 65 of 74 -
Page 66 of 74 -
Page 67 of 74 -
Page 68 of 74 -
Page 69 of 74 -
Page 70 of 74 -
Page 71 of 74 -
Page 72 of 74 -
Page 73 of 74 -
Page 74 of 74