JSONDecodeError while using HuggingFace Inference API with LangChain for Embeddings

I’m trying to generate embeddings using the Hugging Face Inference API with LangChain in Python, but I’m running into issues. My goal is to use the API (not local models) to generate embeddings for text, specifically with the sentence-transformers/all-MiniLM-L6-v2 model. I’m using langchain-huggingface==0.2.0 and huggingface-hub==0.31.2 in a Conda environment with Python 3.10. Here’s the context: I have a valid Hugging Face API token set as an environment variable (HUGGINGFACEHUB_API_TOKEN), and I want to integrate this with LangChain for embeddings.

from langchain_community.embeddings import HuggingFaceInferenceAPIEmbeddings
import os

os.environ["HUGGINGFACEHUB_API_TOKEN"] = "hf_MyTokenHere"  # Valid token confirmed

embeddings = HuggingFaceInferenceAPIEmbeddings(
    api_key=os.environ["HUGGINGFACEHUB_API_TOKEN"],
    model_name="sentence-transformers/all-MiniLM-L6-v2"
)

# Fails here
vector = embeddings.embed_query("Test query")

Error:

---------------------------------------------------------------------------
JSONDecodeError                           Traceback (most recent call last)
File d:\Python_Env\LangChain\venv\lib\site-packages\requests\models.py:974, in Response.json(self, **kwargs)
    973 try:
--> 974     return complexjson.loads(self.text, **kwargs)
    975 except JSONDecodeError as e:
    976     # Catch JSON-related errors and raise as requests.JSONDecodeError
    977     # This aliases json.JSONDecodeError and simplejson.JSONDecodeError

File d:\Python_Env\LangChain\venv\lib\json\__init__.py:346, in loads(s, cls, object_hook, parse_float, parse_int, parse_constant, object_pairs_hook, **kw)
    343 if (cls is None and object_hook is None and
    344         parse_int is None and parse_float is None and
    345         parse_constant is None and object_pairs_hook is None and not kw):
--> 346     return _default_decoder.decode(s)
    347 if cls is None:

File d:\Python_Env\LangChain\venv\lib\json\decoder.py:337, in JSONDecoder.decode(self, s, _w)
    333 """Return the Python representation of ``s`` (a ``str`` instance
    334 containing a JSON document).
    335 
    336 """
--> 337 obj, end = self.raw_decode(s, idx=_w(s, 0).end())
    338 end = _w(s, end).end()

File d:\Python_Env\LangChain\venv\lib\json\decoder.py:355, in JSONDecoder.raw_decode(self, s, idx)
    354 except StopIteration as err:
--> 355     raise JSONDecodeError("Expecting value", s, err.value) from None
    356 return obj, end

JSONDecodeError: Expecting value: line 1 column 1 (char 0)

During handling of the above exception, another exception occurred:

JSONDecodeError                           Traceback (most recent call last)
Cell In[12], line 16
     14 text = "LangChain makes LLM applications modular"
     15 #query_embedding = embeddings.embed_query(text)
---> 16 document_embeddings = embeddings.embed_documents([text, "HuggingFace provides great models"])

File d:\Python_Env\LangChain\venv\lib\site-packages\langchain_community\embeddings\huggingface.py:472, in HuggingFaceInferenceAPIEmbeddings.embed_documents(self, texts)
    441 """Get the embeddings for a list of texts.
    442 
    443 Args:
   (...)
    462         hf_embeddings.embed_documents(texts)
    463 """  # noqa: E501
    464 response = requests.post(
    465     self._api_url,
    466     headers=self._headers,
   (...)
    470     },
    471 )
--> 472 return response.json()

File d:\Python_Env\LangChain\venv\lib\site-packages\requests\models.py:978, in Response.json(self, **kwargs)
    974     return complexjson.loads(self.text, **kwargs)
    975 except JSONDecodeError as e:
    976     # Catch JSON-related errors and raise as requests.JSONDecodeError
    977     # This aliases json.JSONDecodeError and simplejson.JSONDecodeError
--> 978     raise RequestsJSONDecodeError(e.msg, e.doc, e.pos)

JSONDecodeError: Expecting value: line 1 column 1 (char 0)

What I’ve Verified:

My Hugging Face token is valid and has Inference API permissions.

The model sentence-transformers/all-MiniLM-L6-v2 is public and accessible via the UI.

Installed latest versions