highperfocused/danswer
1
0
Fork 0
mirror of https://github.com/danswer-ai/danswer.git synced 2025-10-10 05:05:34 +02:00
Code Issues Packages Projects Releases Wiki Activity

Fix rebase issue with UI-based LLM selection

Browse Source
This commit is contained in:
Weves
2024-04-27 12:51:52 -07:00
committed by Chris Weaver
parent d7a704c0d9
commit 719fc06604
1 changed files with 6 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
backend/ee/danswer/server/query_and_chat/query_backend.py
View File

@@ -10,7 +10,8 @@ from danswer.db.models import User
from danswer.llm.answering.prompts.citations_prompt import ( from danswer.llm.answering.prompts.citations_prompt import (
compute_max_document_tokens_for_persona, compute_max_document_tokens_for_persona,
) )
from danswer.llm.utils import get_default_llm_version from danswer.llm.factory import get_default_llm
from danswer.llm.factory import get_llm_for_persona
from danswer.llm.utils import get_max_input_tokens from danswer.llm.utils import get_max_input_tokens
from danswer.one_shot_answer.answer_question import get_search_answer from danswer.one_shot_answer.answer_question import get_search_answer
from danswer.one_shot_answer.models import DirectQARequest from danswer.one_shot_answer.models import DirectQARequest
@@ -85,11 +86,10 @@ def get_answer_with_quote(
db_session=db_session, db_session=db_session,
) )
llm_name = get_default_llm_version()[0] llm = get_default_llm() if not persona else get_llm_for_persona(persona)
if persona and persona.llm_model_version_override: input_tokens = get_max_input_tokens(
llm_name = persona.llm_model_version_override model_name=llm.config.model_name, model_provider=llm.config.model_provider
)
input_tokens = get_max_input_tokens(model_name=llm_name)
max_history_tokens = int(input_tokens * DANSWER_BOT_TARGET_CHUNK_PERCENTAGE) max_history_tokens = int(input_tokens * DANSWER_BOT_TARGET_CHUNK_PERCENTAGE)
remaining_tokens = input_tokens - max_history_tokens remaining_tokens = input_tokens - max_history_tokens