Chroma db not working in both persistent and http client modes

Question

I am trying to use chroma vector store in Python but I cannot make it work.

When I tried to create a persistent client using the command below I got the below error.

client = chromadb.PersistentClient(path=path="/path/to/save/to")

ERROR: RuntimeError: Chroma is running in http-only client mode, and can only be run with 'chromadb.api.fastapi.FastAPI' as the chroma_api_impl. see https://docs.trychroma.com/usage-guide?lang=py#using-the-python-http-only-client for more information.

I also tried to run chroma in http-only client mode. I used the code below:

from chromadb import HttpClient

from chromadb.config import Settings

settings = Settings(chroma_api_impl="chromadb.api.fastapi.FastAPI", allow_reset=True, anonymized_telemetry=False)

client = HttpClient(host='localhost',port=8000,settings=settings)

it worked but when I tried to create a collection I got the following error:

ERROR: ConnectionError: ('Connection aborted.', RemoteDisconnected('Remote end closed connection without response'))

Any help would be truly appreciated :).

Many many thanks

j3ffyang · Accepted Answer · 2023-12-12 16:03:40Z

2

To create a local non-persistent (data gone after execution finished) Chroma database, you can do

# embedding model as example
embedding_function = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")

# load it into Chroma
db = Chroma.from_documents(docs, embedding_function)

If you want to create/ load from a local persistent Chroma vectorstore,

# save to disk
db2 = Chroma.from_documents(docs, embedding_function,
                            persist_directory="./chroma_db")
docs = db2.similarity_search(query)

# load from disk
db3 = Chroma(persist_directory="./chroma_db", embedding_function=embedding_function)
docs = db3.similarity_search(query)
print(docs[0].page_content)

If you want to pass a Chroma client into LangChain, you would have to have a standalone Chroma vectorstore engine running over HTTP.

Here's the reference doc > https://python.langchain.com/docs/integrations/vectorstores/chroma

answered Dec 12, 2023 at 16:03

j3ffyang

2,52827 silver badges16 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Luca Melli Over a year ago

Many thanks! But when I tried your suggestion I got the below error: ERROR: RuntimeError: Chroma is running in http-only client mode, and can only be run with 'chromadb.api.fastapi.FastAPI' as the chroma_api_impl. see docs.trychroma.com/… for more information. Any other suggestions?

cottontail Over a year ago

this answer works locally but I think OP wanted to get it running over http. I also arrived at this QA for that reason. Anyway, the reference doc link is helpful, but you may want to quote the relevant parts from the docs into your answer as link rot is very likely for these fast evolving libraries.

Aleks G · Accepted Answer · 2024-06-10 16:03:07Z

1

I came here for the same reason. You can solve it with (basically)

model_kwargs=[dict]trust_remote_code=True

edited Jun 10, 2024 at 16:03

Aleks G

57.5k34 gold badges179 silver badges280 bronze badges

answered Jun 4, 2024 at 11:48

Tonic

111 bronze badge

Comments

j3ffyang · Accepted Answer · 2023-12-18 16:10:59Z

0

Rewrite the simple code and please give a try

from langchain.vectorstores import Chroma
from langchain.embeddings import HuggingFaceEmbeddings

# using open source llm and download to local disk
embedding_function = HuggingFaceEmbeddings(
    model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")

# use `from_texts` instead of `from_documents`, for reference
vectorstore = Chroma.from_texts(["harry potter's owl is in the castle."], 
                        embedding_function, persist_directory="./chroma_db")
query = "where is Harry Potter's Hedwig?"
docs = vectorstore.similarity_search(query)
print(docs[0].page_content)

answered Dec 18, 2023 at 16:10

j3ffyang

2,52827 silver badges16 bronze badges

Collectives™ on Stack Overflow

Chroma db not working in both persistent and http client modes

3 Answers 3

2 Comments

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

2 Comments

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related