highperfocused/danswer
1
0
Fork 0
mirror of https://github.com/danswer-ai/danswer.git synced 2025-07-08 05:30:24 +02:00
Code Issues Packages Projects Releases Wiki Activity

Enable default quantization (#4815)

Browse Source 
* Adjust migration

* update default in form

* Add cloud indices for bfloat16

* Update backend/shared_configs/configs.py

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>

* Update vespa schema gen script

* Move embedding configs

* Remove unused imports

* remove import from shared configs

* Remove unused model

---------

Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
This commit is contained in:
Chris Weaver
2025-06-05 14:02:08 -07:00
committed by GitHub
parent 85eeb21b77
commit dc542fd7fa
8 changed files with 229 additions and 223 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
backend/scripts/debugging/onyx_vespa_schemas.py
View File

@ -4,21 +4,26 @@ import argparse
import jinja2
from onyx.configs.embedding_configs import SUPPORTED_EMBEDDING_MODELS
from onyx.db.enums import EmbeddingPrecision
from onyx.utils.logger import setup_logger
from shared_configs.configs import SUPPORTED_EMBEDDING_MODELS
logger = setup_logger()
def write_schema(index_name: str, dim: int, template: jinja2.Template) -> None:
def write_schema(
index_name: str,
dim: int,
embedding_precision: EmbeddingPrecision,
template: jinja2.Template,
) -> None:
index_filename = index_name + ".sd"
schema = template.render(
multi_tenant=True,
schema_name=index_name,
dim=dim,
embedding_precision=EmbeddingPrecision.FLOAT.value,
embedding_precision=embedding_precision.value,
)
with open(index_filename, "w", encoding="utf-8") as f:
@ -41,8 +46,13 @@ def main() -> None:
num_indexes = 0
for model in SUPPORTED_EMBEDDING_MODELS:
write_schema(model.index_name, model.dim, template)
write_schema(model.index_name + "__danswer_alt_index", model.dim, template)
write_schema(model.index_name, model.dim, model.embedding_precision, template)
write_schema(
model.index_name + "__danswer_alt_index",
model.dim,
model.embedding_precision,
template,
)
num_indexes += 2
logger.info(f"Wrote {num_indexes} indexes.")