Add `/auto` behind a feature flag that's disabled for now, even for
staff.

We've decided on a different design for context inference, but there are
parts of /auto that will be useful for that, so we want them in the code
base even if they're unused for now.

Release Notes:

- N/A

---------

Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Marshall Bowers <elliott.codes@gmail.com>
This commit is contained in:
Richard Feldman 2024-09-13 13:17:49 -04:00 committed by GitHub
parent 93a3e8bc94
commit 91ffa02e2c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
42 changed files with 2776 additions and 1054 deletions

View file

@ -1,5 +1,5 @@
use super::*;
use sea_orm::QueryOrder;
use sea_orm::{sea_query::OnConflict, QueryOrder};
use std::str::FromStr;
use strum::IntoEnumIterator as _;
@ -99,6 +99,17 @@ impl LlmDatabase {
..Default::default()
}
}))
.on_conflict(
OnConflict::columns([model::Column::ProviderId, model::Column::Name])
.update_columns([
model::Column::MaxRequestsPerMinute,
model::Column::MaxTokensPerMinute,
model::Column::MaxTokensPerDay,
model::Column::PricePerMillionInputTokens,
model::Column::PricePerMillionOutputTokens,
])
.to_owned(),
)
.exec_without_returning(&*tx)
.await?;
Ok(())

View file

@ -40,6 +40,15 @@ pub async fn seed_database(_config: &Config, db: &mut LlmDatabase, _force: bool)
price_per_million_input_tokens: 25, // $0.25/MTok
price_per_million_output_tokens: 125, // $1.25/MTok
},
ModelParams {
provider: LanguageModelProvider::Zed,
name: "Qwen/Qwen2-7B-Instruct".into(),
max_requests_per_minute: 5,
max_tokens_per_minute: 25_000, // These are arbitrary limits we've set to cap costs; we control this number
max_tokens_per_day: 300_000,
price_per_million_input_tokens: 25,
price_per_million_output_tokens: 125,
},
])
.await
}