| { | |
| "mic": "0.3", | |
| "name": "Whisper assisted speech recognition", | |
| "usage": "WhisperProcessor + WhisperForConditionalGeneration.generate, speech with an encoder-decoder assistant", | |
| "architecture": "WhisperForConditionalGeneration", | |
| "checkpoints": { | |
| "openai/whisper-tiny": { | |
| "fixture": "hf-internal-testing/mic-whisper-tiny-slim" | |
| } | |
| }, | |
| "fixtures": { | |
| "hf-internal-testing/mic-whisper-tiny-slim": { | |
| "built_from": "openai/whisper-tiny", | |
| "covers": [ | |
| "openai/whisper-tiny" | |
| ], | |
| "build": { | |
| "seed": 17 | |
| } | |
| } | |
| }, | |
| "integrations": { | |
| "transformers": { | |
| "entry_point": "integrations/transformers/usage.md", | |
| "format": "runnable-markdown", | |
| "requires": [ | |
| "transformers>=5", | |
| "torch", | |
| "soundfile" | |
| ], | |
| "targets": [ | |
| "checkpoints" | |
| ], | |
| "unsupported": { | |
| "fixtures": "Checkpoint-only speech coverage: neither slim weight kind exposes the assisted-suppression regression #48108 on this input." | |
| }, | |
| "variants": { | |
| "default": {}, | |
| "eager": { | |
| "attn_implementation": "eager" | |
| }, | |
| "bfloat16": { | |
| "dtype": "bfloat16" | |
| } | |
| }, | |
| "training": false, | |
| "resources": { | |
| "openai/whisper-tiny": { | |
| "memory_gb": 2, | |
| "download_mb": 151 | |
| } | |
| } | |
| } | |
| }, | |
| "outputs": { | |
| "runtime": { | |
| "compare": "exact" | |
| }, | |
| "speech": { | |
| "compare": "exact" | |
| }, | |
| "greedy_tokens": { | |
| "compare": "exact" | |
| }, | |
| "greedy_text": { | |
| "compare": "exact" | |
| }, | |
| "assisted_tokens": { | |
| "compare": "exact" | |
| }, | |
| "assisted_text": { | |
| "compare": "exact" | |
| }, | |
| "assisted_matches_greedy": { | |
| "compare": "true" | |
| } | |
| } | |
| } | |