Four safetensors shards of openai/whisper-small (MIT), used to keep each Docker layer below the Vercel Container Registry 2 GB limit. No fine-tuning is stored here; the VoiceBridge LoRA adapter is separate.
model-00004-of-00004.safetensors 246 MBmodel-00001-of-00004.safetensors 241 MBmodel-00002-of-00004.safetensors 241 MBmodel-00003-of-00004.safetensors 239 MB