docs: provider parameter branch and import fixing

[dudududukim]

Problem

1. Current examples throw a 404 Client Error when using HuggingFaceInferenceAPI:

HfHubHTTPError: 404 Client Error: Not Found for url: https://router.huggingface.co/hf-inference/models/Qwen/Qwen2.5-Coder-32B-Instruct/v1/chat/completions

See error examples in my repository notebooks:

• dudududukim/hf_agent/blob/main/llamaindex.ipynb
• dudududukim/hf_agent/blob/main/agent_RAG/main.ipynb

This happens because Hugging Face discontinued free serverless inference for most models in 2024. The model now requires a provider parameter.

2. DuckDuckGoSearchRun was incorrectly imported from tools module instead of langchain_community.tools

Changes

1.

Added provider="auto" to HuggingFaceInferenceAPI in:

• units/en/unit2/llama-index/components.mdx
• units/en/unit3/agentic-rag/agent.mdx

Using provider="auto" lets users configure their preferred providers at https://huggingface.co/settings/inference-providers

2.

Changed import to from langchain_community.tools import DuckDuckGoSearchRun for correct initialization

References

• Docs: https://huggingface.co/docs/inference-providers/
• Available models: https://huggingface.co/inference/models

[dudududukim]

Dear maintainer/reviewer, could you please approve the workflow run? Thank you!