Grading fails with NameResolutionError to api-inference.huggingface.co — all correct answers marked wrong

A grading system for a Code Quiz in Chapter 11 (Fine-Tune Large Language Models) is failing with a NameResolutionError to api-inference.huggingface.co, marking all correct answers wrong. The grader depends on a live call to Qwen/Qwen2.5-Coder-32B-Instruct via the Hugging Face Inference API, and when the DNS resolution fails, it falls back to a strict comparison that marks all answers incorrect.

Hi, I’ve been attempting the Code Quiz in Chapter 11 Fine-Tune Large Language Models and consistently getting a 0% score despite submitting correct solutions. This has happened across multiple attempts on different days. The grader appears to depend on a live call to Qwen/Qwen2.5-Coder-32B-Instruct via the HF Inference API for evaluation, and that call is failing with a DNS resolution error. When it fails, it seems to fall back to a strict comparison that marks all answers incorrect regardless of correctness. Error message from two separate attempts : Error checking code: MaxRetryError ‘HTTPSConnectionPool host=‘ api-inference.huggingface.co http://api-inference.huggingface.co ’, port=443 : Max retries exceeded with url: /models/Qwen/Qwen2.5-Coder-32B-Instruct/v1/chat/completions Caused by NameResolutionError “: Failed to resolve ‘ api-inference.huggingface.co http://api-inference.huggingface.co ’ Errno -5 No address associated with hostname ” ’ Request IDs for reference: