Researchers fine-tuned a Mistral large language model to detect racist covenants in real estate deeds. They also worked with Santa Clara County, CA to develop responsible usage, “including legal review and the creation of a historical registry.”
The research team says this was massively more efficient than manual review and cheaper than using a closed model. They've released the #LLM “to assist the hundreds of jurisdictions engaged in similar efforts.”
https://reglab.github.io/racialcovenants/
#GenAI
Sharon Machlis /
npub1rs…julvp
2024-11-09 13:08:38
Author Public Key
npub1rsnd0ksgq9hfvquvmcxs3vm8pv375nfru7vq98gwwnm9svhcp0ss0julvpPublished at
2024-11-09 13:08:38Event JSON
{
"id": "89e2aa8f21c33528e753b0325ea8a35265e8902f37e9f7998db753eec313be46",
"pubkey": "1c26d7da08016e96038cde0d08b3670b23ea4d23e798029d0e74f65832f80be1",
"created_at": 1731157718,
"kind": 1,
"tags": [
[
"t",
"llm"
],
[
"t",
"genai"
],
[
"proxy",
"https://masto.machlis.com/users/smach/statuses/113453152223187129",
"activitypub"
]
],
"content": "Researchers fine-tuned a Mistral large language model to detect racist covenants in real estate deeds. They also worked with Santa Clara County, CA to develop responsible usage, “including legal review and the creation of a historical registry.”\nThe research team says this was massively more efficient than manual review and cheaper than using a closed model. They've released the #LLM “to assist the hundreds of jurisdictions engaged in similar efforts.”\n\nhttps://reglab.github.io/racialcovenants/\n\n#GenAI",
"sig": "b800b8f6c3a7e875feb4ec9f01d1ce01ac6deeca8837f8c4acc999002cf4da00ca27836da276e704ae9845d804dbe55543a278938ca9fb37b1943ee18c9075c9"
}