Name		Name	Last commit message	Last commit date
parent directory ..
README.md		README.md
Use cases and applications of LLMs - Lecture 6.pdf		Use cases and applications of LLMs - Lecture 6.pdf
query_tables.py		query_tables.py
requirements_short.txt		requirements_short.txt
synthetic_data.py		synthetic_data.py
web_search_queries.csv		web_search_queries.csv

README.md

Use cases and applications of LLMs

Query tables in natural language. See query_tables.py

a. Improve code by loading all sections to avoid hard-coding a section with tables

b. Test the capabilities of reasoning with table data: can it sum up numbers or do some other calculation?
Generate synthetic query with misspellings - only task description is available you need to write the code in synthetic_data.py

a. Find the queries in web_search_queries.csv, containing example queries from different topics and use cases, like map search, job search, travel and tourism, general knowledge and learning

b. Implement code, that will load one query at a time and generate up to N misspellings.

c. Improve robustness of the method, for example, skip known abbreviations, like JFK.

d. Make misspellings capture a variety of error types (phonetic, omission, transposition, repetition)?

e. Test the resulting query variants with your favourite web search engine. Are results equal for the same for all variants of a given query? If not, why do you think this is happening?