Langchain Chroma bug when add documents

[Spycsh]

Describe the bug

Hi, I'm using the latest huggingface_hub 0.28.1 and it will crash and show messy embedding output when doing vectordb.add_documents. Fallback to 0.27.1 can fix the issue.

Could you have a look on this issue, thanks!

Reproduction

1. Start a TEI server

model=BAAI/bge-base-en-v1.5
volume=$PWD/data
docker run -d -p 6060:80 -v $volume:/data -e http_proxy=$http_proxy -e https_proxy=$https_proxy --pull always ghcr.io/huggingface/text-embeddings-inference:cpu-1.5 --model-id $model --auto-truncate

2. Run the code below with python test_chroma_add_documents.py > err.log 2>&1 (The code basically uses that TEI server as the embdding function of Chroma, and do an simple add_document)

from langchain_community.vectorstores import Chroma
from langchain_huggingface import HuggingFaceEndpointEmbeddings
from langchain.schema import Document
import random

# Function to generate random documents
def generate_random_documents(num_documents=10, doc_length=50):
    documents = []
    for _ in range(num_documents):
        text = ''.join(random.choices('abcdefghijklmnopqrstuvwxyz ', k=doc_length))
        documents.append(Document(page_content=text))
    return documents

# Generate random documents
random_documents = generate_random_documents()

tei_embedding_endpoint = "http://localhost:6060"

vector_db = Chroma(
    embedding_function=HuggingFaceEndpointEmbeddings(model=tei_embedding_endpoint),
)

vector_db.add_documents(random_documents)

Logs

The above code succeeded when I installed huggingface-hub==0.27.1, but failed when installing huggingface-hub==0.28.1. The error log seems to print huge embeddings, but with no explicit errors... And when I save the log in a txt file, I find the error log start as following:


/home/sdp/sihanche/test_chroma_add_documents.py:19: LangChainDeprecationWarning: The class `Chroma` was deprecated in LangChain 0.2.9 and will be removed in 1.0. An updated version of the class exists in the :class:`~langchain-chroma package and should be used instead. To use it run `pip install -U :class:`~langchain-chroma` and import as `from :class:`~langchain_chroma import Chroma``.
  vector_db = Chroma(
/root/sihanche_venv/lib/python3.12/site-packages/huggingface_hub/utils/_deprecation.py:131: FutureWarning: 'post' (from 'huggingface_hub.inference._client') is deprecated and will be removed from version '0.31.0'. Making direct POST requests to the inference server is not supported anymore. Please use task methods instead (e.g. `InferenceClient.chat_completion`). If your use case is not supported, please open an issue in https://github.com/huggingface/huggingface_hub.
  warnings.warn(warning_message, FutureWarning)
Traceback (most recent call last):
  File "/root/sihanche_venv/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 90, in wrapper
    return func(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/sihanche_venv/lib/python3.12/site-packages/chromadb/api/models/CollectionCommon.py", line 389, in _validate_and_prepare_upsert_request
    upsert_records = normalize_insert_record_set(
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/sihanche_venv/lib/python3.12/site-packages/chromadb/api/types.py", line 187, in normalize_insert_record_set
    base_record_set = normalize_base_record_set(
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/sihanche_venv/lib/python3.12/site-packages/chromadb/api/types.py", line 164, in normalize_base_record_set
    embeddings=normalize_embeddings(embeddings),
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/sihanche_venv/lib/python3.12/site-packages/chromadb/api/types.py", line 88, in normalize_embeddings
    raise ValueError(
ValueError: Expected embeddings to be a list of floats or ints, a list of lists, a numpy array, or a list of numpy arrays, got [[[[3.005876302719116, 2.1381373405456543, 0.9481216669082642, 8.15536880493164, 1.4738448858261108, 2.220341682434082, 0.04728309437632561, -1.8110272884368896, -0.5441240668296814, 0.32091569900512695, 1.4771777391433716, 1.627296805381775, 0.9266064167022705, 1.2391126155853271, -0.48665082454681396, -1.1952226161956787, -0.4621836245059967, 2.351332187652588, 0.7196410298347473, -2.577847480773926, 0.8533875346183777, -3.2535433769226074, 1.9740791320800781, 1.375751256942749, 1.3318135738372803, -4.031718730926514, 2.391464948654175, -39.04726028442383, 2.434889078140259, -1.9554494619369507, 1.0033003091812134, -0.6465897560119629, 1.0250786542892456, -1.0325400829315186, -2.477579355239868, 1.0192513465881348, 0.386062353849411, -0.4952068030834198, 1.8315757513046265, -0.8859990835189819, -0.4211207926273346, -2.7653861045837402, -0.8620356321334839, -1.4569767713546753, -1.4128832817077637, 2.05782413482666, -0.11006411910057068, -0.5385697484016418, -1.4141700267791748, 0.5901338458061218, 1.840973138809204, 1.2116774320602417, -2.1155853271484375, -0.9022896885871887, -3.381002187728882, -0.22921833395957947, 1.783563494682312, -1.6340497732162476, 1.26601243019104, 1.12820303440094, 0.825856626033783, 1.2516766786575317, 1.4689359664916992, -0.441506952047348, -2.6451971530914307, -0.11896096169948578, -3.4909539222717285, -1.255056619644165, 0.4156877398490906, 0.9307308793067932, 0.703662633895874, 0.03833493962883949, 1.3183388710021973, -0.1265009045600891, -2.537118434906006, -0.14639760553836823, 1.2940855026245117, -0.6394643187522888, -1.9029948711395264, 0.6988447308540344, -0.4429458975791931, 2.89530611038208, -3.234011650085449, -0.980390191078186, 0.9419354200363159, 1.2371937036514282, 3.685396432876587, 0.49835142493247986, -0.5340624451637268, -3.2260754108428955, 0.3335322439670563, 1.23859703540802, -0.71039217710495, -0.1361103355884552, -0.24770885705947876, 0.7179780006408691, -1.8748948574066162, 1.3112727403640747, -2.4263319969177246, -0.7393266558647156, -0.7231326103210449, -2.8478591442108154, -2.774787187576294, -1.7655466794967651, 2.1261065006256104, 0.43441200256347656, 0.6733825206756592, -0.6575789451599121, -0.4900713860988617, -1.5264394283294678, 2.433136224746704, -1.4786337614059448, -0.07689659297466278, 0.3222745358943939, -0.7179399728775024, -1.0739233493804932, 10.620174407958984, -3.5914392471313477, 0.15320725739002228, 0.1542065292596817, 1.4247353076934814, 0.4752683639526367, 0.08139839768409729, 3.1361100673675537, 0.09998984634876251, -0.7244051098823547, -0.4922061860561371, 1.100483775138855, -1.5067870616912842, -1.0973107814788818, 0.12099336832761765, 1.9025170803070068, 1.7372387647628784, 1.867568016052246, -1.808851718902588, -0.004878419451415539, -0.5992695093154907, 2.5557920932769775, 0.6204664707183838, -0.6507046818733215, 1.3751416206359863, -0.

System info

- huggingface_hub version: 0.28.1
- Platform: Linux-6.8.0-50-generic-x86_64-with-glibc2.39
- Python version: 3.12.3
- Running in iPython ?: No
- Running in notebook ?: No
- Running in Google Colab ?: No
- Running in Google Colab Enterprise ?: No
- Token path ?: /root/.cache/huggingface/token
- Has saved token ?: False
- Configured git credential helpers: 
- FastAI: N/A
- Tensorflow: N/A
- Torch: 2.6.0
- Jinja2: 3.1.5
- Graphviz: N/A
- keras: N/A
- Pydot: N/A
- Pillow: 11.1.0
- hf_transfer: N/A
- gradio: N/A
- tensorboard: N/A
- numpy: 2.2.1
- pydantic: 2.10.5
- aiohttp: 3.11.11
- ENDPOINT: https://huggingface.co
- HF_HUB_CACHE: /root/.cache/huggingface/hub
- HF_ASSETS_CACHE: /root/.cache/huggingface/assets
- HF_TOKEN_PATH: /root/.cache/huggingface/token
- HF_STORED_TOKENS_PATH: /root/.cache/huggingface/stored_tokens
- HF_HUB_OFFLINE: False
- HF_HUB_DISABLE_TELEMETRY: False
- HF_HUB_DISABLE_PROGRESS_BARS: None
- HF_HUB_DISABLE_SYMLINKS_WARNING: False
- HF_HUB_DISABLE_EXPERIMENTAL_WARNING: False
- HF_HUB_DISABLE_IMPLICIT_TOKEN: False
- HF_HUB_ENABLE_HF_TRANSFER: False
- HF_HUB_ETAG_TIMEOUT: 10
- HF_HUB_DOWNLOAD_TIMEOUT: 10

[julien-c]

We might need to open a PR to update langchain_huggingface.HuggingFaceEndpointEmbeddings (and maybe ping a recent huggingface_hub in there)