Embeddings hardware recommendations¶
BAAI/bge-m3 served through TEI for Easy8 AI retrieval runs on NVIDIA GPU infrastructure only. CPU-only deployment is not supported.
Requirements¶
| Item | Requirement |
|---|---|
| GPU | 1 NVIDIA GPU |
| VRAM | 4 GB minimum, 8 GB or more recommended |
| Host | NVIDIA drivers installed |
| Container runtime | Docker with the NVIDIA Container Toolkit configured for GPU access in containers |
Notes¶
EMBEDDINGS_URLpoints to the TEI endpoint, see PRO and Helpdesk with self-hosted AI.- More VRAM allows larger batches and faster indexing.
- TEI can share a GPU with vLLM, see LLM hardware recommendations.