AI/dify
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/langgenius/dify.git synced 2025-08-12 02:19:05 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat: decrease claude model request time by set max top_k to 10 (#1071)

Browse Source
This commit is contained in:
takatost 2023-08-31 18:23:44 +08:00 committed by GitHub
parent 9458b8978f
commit fcc6b41ab7
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
api/core/orchestrator_rule_parser.py
View File

@ -283,6 +283,7 @@ class OrchestratorRuleParser:
def _dynamic_calc_retrieve_k(cls, dataset: Dataset, rest_tokens: int) -> int: def _dynamic_calc_retrieve_k(cls, dataset: Dataset, rest_tokens: int) -> int:
DEFAULT_K = 2 DEFAULT_K = 2
CONTEXT_TOKENS_PERCENT = 0.3 CONTEXT_TOKENS_PERCENT = 0.3
MAX_K = 10
if rest_tokens == -1: if rest_tokens == -1:
return DEFAULT_K return DEFAULT_K
@ -311,5 +312,5 @@ class OrchestratorRuleParser:
if context_limit_tokens <= segment_max_tokens * DEFAULT_K: if context_limit_tokens <= segment_max_tokens * DEFAULT_K:
return DEFAULT_K return DEFAULT_K
# Expand the k value when there's still some room left in the 30% rest tokens space # Expand the k value when there's still some room left in the 30% rest tokens space, but less than the MAX_K
return context_limit_tokens // segment_max_tokens return min(context_limit_tokens // segment_max_tokens, MAX_K)