We lifted 2 functions out of this paper's own repositories and ran 2 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| huihanlhh/culture-gen | canonical | 2 of 2 |
| Function | Status | Where it lives |
|---|---|---|
| extract_keyword_probability_distribution | Ran | huihanlhh/culture-gen/script/culture_symbols.py code served (permissive licence) · get_code("9e998932b0ab4052") |
| simpson_iod | Ran | huihanlhh/culture-gen/script/cultural_evaluations.py code served (permissive licence) · get_code("6221576fbe8a2441") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
As the utilization of large language models (LLMs) has proliferated world-wide, it is crucial for them to have adequate knowledge and fair representation for diverse global cultures. In this work, we uncover culture perceptions of three SOTA models on 110 countries and regions on 8 culture-related topics through culture-conditioned generations, and extract symbols from these generations that are associated to each culture by the LLM. We discover that culture-conditioned generation consist of linguistic "markers" that distinguish marginalized cultures apart from default cultures. We also discover that LLMs have an uneven degree of diversity in the culture symbols, and that cultures from different geographic regions have different presence in LLMs' culture-agnostic generation. Our findings promote further research in studying the knowledge and fairness of global culture perception in LLMs. Code and Data can be found here: https://github.com/huihanlhh/Culture-Gen/
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2404.10199")
get_code_for_paper("2404.10199")
have("2404.10199")
Connect an agent — have() is free.