HuggingFaceEndpointEmbeddings fails with faiss

[tocab]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

This code fails:

from langchain_huggingface.embeddings import HuggingFaceEndpointEmbeddings
from langchain_community.vectorstores import FAISS
import numpy as np

hf_embeddings = HuggingFaceEndpointEmbeddings(
    model="deepseek-ai/DeepSeek-R1-Distill-Qwen-32B",
    task="feature-extraction",
    huggingfacehub_api_token=inference_token,
)
vector = FAISS.from_texts(["Hello, how are you?", "I am fine, thank you."], hf_embeddings)

Most likely because of this:

embeddings = hf_embeddings.embed_documents(["Hello, how are you?", "I am fine, thank you."])
np.array(embeddings)

Both returns the error:
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 2 dimensions. The detected shape was (2, 1) + inhomogeneous part.

Error Message and Stack Trace (if applicable)

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[5], line 11
      6 hf_embeddings = HuggingFaceEndpointEmbeddings(
      7     model="deepseek-ai/DeepSeek-R1-Distill-Qwen-32B",
      8     task="feature-extraction",
      9     huggingfacehub_api_token=inference_token,
     10 )
---> 11 vector = FAISS.from_texts(["Hello, how are you?", "I am fine, thank you."], hf_embeddings)

/.venv/lib/python3.12/site-packages/langchain_community/vectorstores/faiss.py:1044, in FAISS.from_texts(cls, texts, embedding, metadatas, ids, **kwargs)
   1025 """Construct FAISS wrapper from raw documents.
   1026 
   1027 This is a user friendly interface that:
   (...)
   1041         faiss = FAISS.from_texts(texts, embeddings)
   1042 """
   1043 embeddings = embedding.embed_documents(texts)
-> 1044 return cls.__from(
   1045     texts,
   1046     embeddings,
   1047     embedding,
   1048     metadatas=metadatas,
   1049     ids=ids,
   1050     **kwargs,
   1051 )

/.venv/lib/python3.12/site-packages/langchain_community/vectorstores/faiss.py:1013, in FAISS.__from(cls, texts, embeddings, embedding, metadatas, ids, normalize_L2, distance_strategy, **kwargs)
   1003 index_to_docstore_id = kwargs.pop("index_to_docstore_id", {})
   1004 vecstore = cls(
   1005     embedding,
   1006     index,
   (...)
   1011     **kwargs,
   1012 )
-> 1013 vecstore.__add(texts, embeddings, metadatas=metadatas, ids=ids)
   1014 return vecstore

/.venv/lib/python3.12/site-packages/langchain_community/vectorstores/faiss.py:310, in FAISS.__add(self, texts, embeddings, metadatas, ids)
    308     raise ValueError("Duplicate ids found in the ids list.")
    309 # Add to the index.
--> 310 vector = np.array(embeddings, dtype=np.float32)
    311 if self._normalize_L2:
    312     faiss.normalize_L2(vector)

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 2 dimensions. The detected shape was (2, 1) + inhomogeneous part.

Description

I tried following the quickstart guide with the huggingface api. I don't know if the same error happens with the OpenAI embeddings endpoint, since I don't have API access there.

System Info

System Information
------------------
> OS:  Linux
> OS Version:  #53~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Jan 15 19:18:46 UTC 2
> Python Version:  3.12.7 (main, Oct 20 2024, 13:47:02) [GCC 11.4.0]

Package Information
-------------------
> langchain_core: 0.3.33
> langchain: 0.3.17
> langchain_community: 0.3.16
> langsmith: 0.3.4
> langchain_huggingface: 0.1.2
> langchain_text_splitters: 0.3.5

Optional packages not installed
-------------------------------
> langserve

Other Dependencies
------------------
> aiohttp: 3.11.11
> async-timeout: Installed. No version info available.
> dataclasses-json: 0.6.7
> httpx: 0.27.2
> httpx-sse: 0.4.0
> huggingface-hub: 0.28.1
> jsonpatch: 1.33
> langsmith-pyo3: Installed. No version info available.
> numpy: 2.2.2
> orjson: 3.10.15
> packaging: 24.2
> pydantic: 2.10.6
> pydantic-settings: 2.7.1
> pytest: 8.3.4
> PyYAML: 6.0.2
> requests: 2.32.3
> requests-toolbelt: 1.0.0
> rich: 13.9.4
> sentence-transformers: 3.4.1
> SQLAlchemy: 2.0.37
> tenacity: 9.0.0
> tokenizers: 0.21.0
> transformers: 4.48.2
> typing-extensions: 4.12.2
> zstandard: 0.23.0