13 lines
1.4 KiB
YAML
13 lines
1.4 KiB
YAML
inference_eval:
|
|
model: Nanthasit/sakthai-context-1.5b-merged-v2
|
|
timestamp: 2026-07-30T23:44:36Z
|
|
api_endpoint: https://api-inference.huggingface.co/models/Nanthasit/sakthai-context-1.5b-merged-v2
|
|
status: FAILED
|
|
error: "DNS resolution failed: api-inference.huggingface.co does not resolve (NXDOMAIN confirmed via multiple DNS servers). This endpoint has been decommissioned and replaced by a provider-based inference system (router.huggingface.co/hf-inference)."
|
|
diagnostics:
|
|
- "Old Inference API endpoint api-inference.huggingface.co: DNS NXDOMAIN (no A/AAAA records)"
|
|
- "New router endpoint router.huggingface.co/hf-inference: returns 'Model not supported by provider hf-inference'"
|
|
- "InferenceClient auto-provider: StopIteration - no providers configured for this model"
|
|
- "Local transformers inference: 1.5B model too large for environment (1.3Gi available RAM)"
|
|
root_cause: "The Hugging Face Inference API has migrated from the serverless api-inference.huggingface.co endpoint to a provider-based system (Inference Providers). Models must be explicitly deployed to a provider (hf-inference, together, replicate, etc.) to be accessible via the API. This model has no provider deployment."
|
|
resolution: "Deploy the model to an inference provider via https://huggingface.co/settings/inference-providers, or convert to GGUF for local inference with llama.cpp" |