KNIME Hugging Face: HF Hub LLM Connector & LLM Prompter can only be green executed when using HuggingFaceH4/zephyr-7b-beta model

KNIME Extensions
1

Workflow
LLMs for Sentiment Prediction
Sentiment Analysis AI GenAI HuggingFace
Roberto Cadili
Versionv1.1
Created on Nov 16, 2024 8:22 AM

image
image1920Ă—1080 309 KB

LLM Prompter can only be successfly executed (green) when
HuggingFaceH4/zephyr-7b-beta is being used and even using bigscience/mt0-large generates error 404.

2

image
image1920Ă—1080 367 KB

Execute failed: 404 Client Error: Not Found for url: https://api-inference.huggingface.co/models/bigscience/mt0-large (Request ID: Root=1-68447f4b-110f79ce00efa05d3be0c0d1;85d2b285-eae7-45d2-b1b1-c526dca1930d)

despite the nodes of Credentials Configuration (Hugging Face API Key) & HF Hub Authenticator & HF Hub LLM Connector can be executed (green) successfully.

I have tried other HF LLM:

but got error 404 in LLM Prompter node output

3

image
image1920Ă—1080 375 KB

I have replaced the default HF Hub LLM connector setting using

with

but still got 404 error message

Execute failed: 404 Client Error: Not Found for url: https://api-inference.huggingface.co/models/llmware/slim-sentiment-tool (Request ID: Root=1-68456f5b-1248a1e571e89053343f942b;702d703c-3455-46ae-b6f8-5940850ef73b)

4

Hi @HMDATA247,

You’re right — the LLM Connector and Prompter nodes in KNIME currently support models that are hosted directly on Hugging Face’s Inference API. That’s why models like HuggingFaceH4/zephyr-7b-beta work, while others like bigscience/mt0-large or deepseek-ai models don’t — they aren’t served through Hugging Face’s own infrastructure.

Look for models that list HF Inference API under the “Inference Providers” section on Hugging Face.

image

There is a ticket to support other Inference Providers in Hugging Face Hub Connectors (AP-24349) for the future.

Best,
Keerthan

2 Likes
5

Thank you very much for your prompt reply and guidance. Highly appreciate it. :pray:

1 Like