[MLsys2026 Best Paper]: https://arxiv.org/abs/2506.08276. RAG on Everything with LEANN. Enjoy 97% storage savings while running a fast, accurate, and 100% private RAG application on your personal device.
Python aifaissgpt-osslangchainllama-indexllmlocalstorageoffline-firstollamaprivacypythonragretrieval-augmented-generationvector-databasevector-searchvectors
fix(server, interactive): resolve pydantic ForwardRef and readline history errors (#304)
Rob Grayf86beeb23aea
diff --git a/packages/leann-core/src/leann/interactive_utils.py b/packages/leann-core/src/leann/interactive_utils.pyindex da05d8b..884e7b7 100644--- a/packages/leann-core/src/leann/interactive_utils.py+++ b/packages/leann-core/src/leann/interactive_utils.py@@ -73,7 +73,7 @@ class InteractiveSession:try:rl.read_history_file(str(history_file))rl.set_history_length(1000)- except FileNotFoundError:+ except (FileNotFoundError, FileExistsError, OSError):pass# Save history on exitdiff --git a/packages/leann-core/src/leann/server.py b/packages/leann-core/src/leann/server.pyindex ecc07a3..2041513 100644--- a/packages/leann-core/src/leann/server.py+++ b/packages/leann-core/src/leann/server.py@@ -19,10 +19,30 @@ import osfrom pathlib import Pathfrom typing import Any+from pydantic import BaseModel as _BaseModel+from .api import LeannSearcherfrom .cli import LeannCLI+class SearchRequest(_BaseModel):+ query: str+ top_k: int = 5+ complexity: int = 64+ beam_width: int = 1+ prune_ratio: float = 0.0+ recompute_embeddings: bool = True+ pruning_strategy: str = "global"+ use_grep: bool = False+++class SearchResultModel(_BaseModel):+ id: str+ score: float+ text: str+ metadata: dict[str, Any]++def _ensure_fastapi():"""Lazy import FastAPI and Pydantic, with a clear error if missing."""try:@@ -95,22 +115,6 @@ def create_app():FastAPI, HTTPException, BaseModel = _ensure_fastapi()- class SearchRequest(BaseModel):- query: str- top_k: int = 5- complexity: int = 64- beam_width: int = 1- prune_ratio: float = 0.0- recompute_embeddings: bool = True- pruning_strategy: str = "global"- use_grep: bool = False-- class SearchResultModel(BaseModel):- id: str- score: float- text: str- metadata: dict[str, Any]-app = FastAPI(title="LEANN Vector DB Server",description=(