air-kyi
commented
5 days ago My kernel crashes with from_documents too.
Example Code
self.embed_model = AzureOpenAIEmbeddings(api_key=self.api_key,
azure_endpoint=self.endpoint,
model="text-embedding-ada-002",
chunk_size=1)
loader = DataFrameLoader(pandas_dataframe, page_content_column='column')
documents = loader.load()
vectorstore = Chroma.from_documents(documents=documents,
embedding=self.embed_model,
persist_directory='my-dir')
vectorstore.persist()Error Message and Stack Trace (if applicable)
11:02:02.230 [info] Generated code for 27 = <ipython-input-27-bf686321592c> with 20 lines
11:02:18.176 [error] Disposing session as kernel process died ExitCode: 3221225477, Reason: Description
I was using Chroma with no issue for months but accidentally deleted Conda and now using pip-only venv. Recreated a new venv and now when using the from_documents method of Chroma from langchain, the Python kernel crashes without any error messages. The process finishes with exit code 3221225477. I am using the latest version of Chroma from langchain_chroma and have tried:
- Running on ipython
- Running in terminal (returns
Segmentation Fault, I think this is overall a memory error) - Downgrading
langchain_community - Using
langchain_communitydeprecated version instead oflangchain_chroma - Getting (
.get) or setting (.from_documents) methods for vectorstore - Using different sized datasets (100, 500)
- Recreating venv
- Using Chromadb instead of Langchain Chroma
System Info
OS: Windows Version 10.0.19045 Build 19045 Python 3.11.1
Package Information
langchain-chroma==0.1.4 ├── chromadb [required: >=0.4.0,<0.6.0,!=0.5.5,!=0.5.4, installed: 0.5.18] ├── fastapi [required: >=0.95.2,<1, installed: 0.115.4] ├── langchain-core [required: >=0.1.40,<0.4, installed: 0.3.15] └── numpy [required: >=1,<2, installed: 1.26.4] langchain-community==0.3.5 ├── aiohttp [required: >=3.8.3,<4.0.0, installed: 3.10.10] ├── dataclasses-json [required: >=0.5.7,<0.7, installed: 0.6.7] ├── httpx-sse [required: >=0.4.0,<0.5.0, installed: 0.4.0] ├── langchain [required: >=0.3.6,<0.4.0, installed: 0.3.7] ├── langchain-core [required: >=0.3.15,<0.4.0, installed: 0.3.15] ├── langsmith [required: >=0.1.125,<0.2.0, installed: 0.1.140] ├── numpy [required: >=1,<2, installed: 1.26.4] ├── pydantic-settings [required: >=2.4.0,<3.0.0, installed: 2.6.1] ├── PyYAML [required: >=5.3, installed: 6.0.2] ├── requests [required: >=2,<3, installed: 2.32.3] ├── SQLAlchemy [required: >=1.4,<2.0.36, installed: 2.0.35] └── tenacity [required: >=8.1.0,<10,!=8.4.0, installed: 8.5.0] langchain-openai==0.2.6 ├── langchain-core [required: >=0.3.15,<0.4.0, installed: 0.3.15] ├── openai [required: >=1.54.0,<2.0.0, installed: 1.54.2] └── tiktoken [required: >=0.7,<1, installed: 0.8.0]
Similar issues
https://github.com/chroma-core/chroma/issues/2513 https://github.com/chroma-core/chroma/issues/3058
Checked other resources
Example Code
from langchain_community.vectorstores import Chroma embed_model_path = '.././AI-ModelScope/bge-small-en-v1___5' from langchain_huggingface import HuggingFaceEmbeddings embedding = HuggingFaceEmbeddings(model_name=embed_model_path) texts = [ "Test" ] try: smalldb_chinese = Chroma.from_texts(texts, embedding=embedding) except Exception as r: print('%s' %(r))
Error Message and Stack Trace (if applicable)
Process finished with exit code -1073741819 (0xC0000005)
Description
When using the from_texts method of Chroma from langchain, the Python kernel crashes without any error messages. The process finishes with exit code -1073741819 (0xC0000005). This issue occurs consistently and makes it impossible to use this method effectively. I am using the latest version of Chroma from langchain and have tried on different environments but still encounter the same problem. Any help or suggestions to resolve this issue would be greatly appreciated.
System Info
System Information
Package Information
Optional packages not installed
Other Dependencies