The state of NLP in production 🥽

NLP in production vs real life

Transcript

1. The state of NLP in production
3. Python Mauritius Usergroup site fb linkedin mailing list 3
4. url pymug.com site 4
5. About me compileralchemy.com 5
6. slides 6
7. The state of NLP in production 7
8. Hardest part of a real-world project 8
9. ? 9
10. Is it cooking up an awesome model? 10
11. No, the world is more complex than this 11
12. Elements of an NLP project 12
13. NLP project gather data clean store train use model retrain model 13
14. gather data 14
15. Toy project use curated data set quick extraction 15
16. Real project a lot of data needed data corresponds to business case. data probably does not exist speed of data gathering find ingenious / better ways of getting data automate collection 16
17. clean/preprocess data 17
18. Toy project use an existing parser / curator e.g. NLTK existing options 18
19. Real project use a parser intended for it, several custom steps parallel processing of data 19
20. store data 20
21. Toy project laptop 21
22. Real project cloud database hot / cold data TTL 22
23. training 23
24. Toy project use laptop / external GPU 24
25. Real project on cloud training on cloud knowledge cross-cloud skills fault tolerance 25
26. use model 26
27. Toy project local website / code 27
28. Real project continuation of pipeline web service architecture devops / deploy 28
29. retraining 29
30. Toy project euhh this even exists???? 30
31. Real project learn cloud offerings for continuous learning ways to retrain / fine tune 31
32. It's more than serving a model 32
33. Operation model 33
34. [ pipeline ] data collection --- process --- train -<- | | --------------------------- model ^ | | | | --->--- V web service [pod] [pod] --- happy user | -> users service [pod] [pod] | -> db service [pod] 34
35. skills chart 35
36. skills --------------- --------------- | | | | | backend | | devops | | | | | --------------- --------------- --------------- --------------- | | | | | backend | | data eng | | | | | --------------- --------------- 36
37. skills --------------- --------------- | | | | | backend | | devops | | | | | --------------- --------------- web service deploy --------------- --------------- | | | | | ml | | data eng | | | | | --------------- --------------- models pipelining 37
38. code blueprint [ architecture repos ] [ pipeline repos ] [ ml repos ] [ backend repos ] 38
39. Tools 39
40. Pandas Good queries Much resources Read SQL 40
41. Dask Good for it's purpose: Parallelize tasks Poor docs 41
42. Polars Awesome parallelizations Great docs 42
43. NLTK use spacy if possible 43
44. Notebooks great for cloud used in production on the cloud 44
45. Advice to research / scientists folks keep everything clean people will come after you always in hurry / messy / i'll clean it later mood good practices? is this phrase in the korean dictionary? 45
46. General advices have great docs good onboarding have great standards 46
47. Keep learning! 47