| { | |
| "architectures": [ | |
| "FunAsrNanoEncoder" | |
| ], | |
| "attention_dropout": 0.1, | |
| "dtype": "float32", | |
| "fsmn_kernel_size": 5, | |
| "hidden_act": "gelu", | |
| "hidden_dropout": 0.0, | |
| "hidden_size": 64, | |
| "intermediate_size": 64, | |
| "layer_norm_eps": 1e-05, | |
| "max_position_embeddings": 200, | |
| "model_type": "fun_asr_nano_encoder", | |
| "num_attention_heads": 4, | |
| "num_hidden_layers": 2, | |
| "num_mel_bins": 80, | |
| "num_stacked_frames": 1, | |
| "num_timestamp_prediction_layers": 1, | |
| "transformers_version": "5.19.0.dev0" | |
| } | |