Skip to content

armgit / LEANN

Public Imported from GitHub
0

[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.

fix: guard against unsupported transformers

Aakash Sureshaa9864c0a957
diff --git a/apps/colqwen_rag.py b/apps/colqwen_rag.py
index ac05250..b4fea09 100644
--- a/apps/colqwen_rag.py
+++ b/apps/colqwen_rag.py
@@ -24,8 +24,6 @@ if str(_leann_hnsw_pkg) not in sys.path:
sys.path.append(str(_leann_hnsw_pkg))
import torch # noqa: E402
-from colpali_engine import ColPali, ColPaliProcessor, ColQwen2, ColQwen2Processor # noqa: E402
-from colpali_engine.utils.torch_utils import ListDataset # noqa: E402
from pdf2image import convert_from_path # noqa: E402
from PIL import Image # noqa: E402
from torch.utils.data import DataLoader # noqa: E402
@@ -46,6 +44,7 @@ class ColQwenRAG:
Args:
model_type: "colqwen2" or "colpali"
"""
+ self._assert_supported_transformers()
self.model_type = model_type
self.device = self._get_device()
# Use float32 on MPS to avoid memory issues, float16 on CUDA, bfloat16 on CPU
@@ -60,6 +59,13 @@ class ColQwenRAG:
# Load model and processor with MPS-optimized settings
try:
+ from colpali_engine import (
+ ColPali,
+ ColPaliProcessor,
+ ColQwen2,
+ ColQwen2Processor,
+ )
+
if model_type == "colqwen2":
self.model_name = "vidore/colqwen2-v1.0"
if self.device.type == "mps":
@@ -123,6 +129,39 @@ class ColQwenRAG:
else:
raise
+ def _assert_supported_transformers(self) -> None:
+ """Fail fast on unsupported transformers versions."""
+ from importlib.metadata import PackageNotFoundError, version
+
+ try:
+ transformers_version = version("transformers")
+ except PackageNotFoundError:
+ return
+
+ def _parse_semver(value: str) -> tuple[int, int, int]:
+ parts = value.split(".")
+ numbers: list[int] = []
+ for part in parts[:3]:
+ digits = []
+ for ch in part:
+ if ch.isdigit():
+ digits.append(ch)
+ else:
+ break
+ numbers.append(int("".join(digits)) if digits else 0)
+ while len(numbers) < 3:
+ numbers.append(0)
+ return tuple(numbers) # type: ignore[return-value]
+
+ if _parse_semver(transformers_version) >= (4, 46, 0):
+ raise RuntimeError(
+ "Unsupported transformers version detected. "
+ "LEANN currently requires transformers<4.46 due to typing changes "
+ "and transformers 5.x removing symbols such as HybridCache. "
+ "Please install a compatible version, e.g. "
+ '`pip install "transformers<4.46"`.'
+ )
+
def _get_device(self):
"""Auto-select best available device."""
if torch.cuda.is_available():
@@ -288,6 +327,8 @@ class ColQwenRAG:
if not images:
raise RuntimeError("No images provided for embedding.")
+ from colpali_engine.utils.torch_utils import ListDataset
+
dataset = ListDataset(images)
dataloader = DataLoader(dataset, batch_size=1, shuffle=False, collate_fn=lambda x: x)