{ "mic": "0.3", "name": "Hubert CTC speech recognition", "usage": "Wav2Vec2Processor + HubertForCTC, batched audio with padding, CTC loss", "architecture": "HubertForCTC", "checkpoints": { "facebook/hubert-large-ls960-ft": { "fixture": "huggingface-mic/hubert-large-ls960-ft-mic" } }, "fixtures": { "huggingface-mic/hubert-large-ls960-ft-mic": { "built_from": "facebook/hubert-large-ls960-ft", "covers": [ "facebook/hubert-large-ls960-ft" ], "why_separate": "Full model has 380M parameters; fixture shrinks to ~100K for fast CI", "build": { "seed": 17, "model_class": "HubertForCTC", "config_overrides": { "fields": { "num_hidden_layers": 2, "hidden_size": 32, "intermediate_size": 64, "num_attention_heads": 4, "conv_dim": [ 32, 32, 32, 32, 32, 32, 32 ], "num_conv_pos_embeddings": 32, "num_conv_pos_embedding_groups": 4 }, "reason": "Shrink transformer and conv dimensions for fast fixture while preserving architecture shape and stride product (320)" } } } }, "integrations": { "transformers": { "entry_point": "integrations/transformers/usage.md", "format": "runnable-markdown", "requires": [ "transformers>=5", "torch", "numpy" ], "targets": [ "fixtures", "checkpoints" ], "variants": { "default": {}, "eager": { "attn_implementation": "eager" }, "bfloat16": { "dtype": "bfloat16" } }, "training": true, "resources": { "fixture": { "memory_gb": 1, "download_mb": 5 }, "facebook/hubert-large-ls960-ft": { "memory_gb": 4, "download_mb": 1500, "variants": { "bfloat16": { "memory_gb": 2 } } } } } }, "outputs": { "runtime": { "compare": "exact" }, "probe_logits": { "compare": "allclose", "atol": 0.0001, "rtol": 0.0001 }, "transcription": { "compare": "exact" }, "training.loss": { "compare": "allclose", "atol": 0.0001, "rtol": 0.0001 }, "training.gradients": { "compare": "all_true" }, "training.updated": { "compare": "true" } } }