RedHatAI/whisper-tiny-FP8-Dynamic

1{2  "_name_or_path": "openai/whisper-tiny",3  "activation_dropout": 0.0,4  "activation_function": "gelu",5  "apply_spec_augment": false,6  "architectures": [7    "WhisperForConditionalGeneration"8  ],9  "attention_dropout": 0.0,10  "begin_suppress_tokens": null,11  "bos_token_id": 50257,12  "classifier_proj_size": 256,13  "d_model": 384,14  "decoder_attention_heads": 6,15  "decoder_ffn_dim": 1536,16  "decoder_layerdrop": 0.0,17  "decoder_layers": 4,18  "decoder_start_token_id": 50258,19  "dropout": 0.0,20  "encoder_attention_heads": 6,21  "encoder_ffn_dim": 1536,22  "encoder_layerdrop": 0.0,23  "encoder_layers": 4,24  "eos_token_id": 50257,25  "forced_decoder_ids": null,26  "init_std": 0.02,27  "is_encoder_decoder": true,28  "mask_feature_length": 10,29  "mask_feature_min_masks": 0,30  "mask_feature_prob": 0.0,31  "mask_time_length": 10,32  "mask_time_min_masks": 2,33  "mask_time_prob": 0.05,34  "max_length": null,35  "max_source_positions": 1500,36  "max_target_positions": 448,37  "median_filter_width": 7,38  "model_type": "whisper",39  "num_hidden_layers": 4,40  "num_mel_bins": 80,41  "pad_token_id": 50257,42  "quantization_config": {43    "config_groups": {44      "group_0": {45        "input_activations": {46          "actorder": null,47          "block_structure": null,48          "dynamic": true,49          "group_size": null,50          "num_bits": 8,51          "observer": null,52          "observer_kwargs": {},53          "strategy": "token",54          "symmetric": true,55          "type": "float"56        },57        "output_activations": null,58        "targets": [59          "Linear"60        ],61        "weights": {62          "actorder": null,63          "block_structure": null,64          "dynamic": false,65          "group_size": null,66          "num_bits": 8,67          "observer": "minmax",68          "observer_kwargs": {},69          "strategy": "channel",70          "symmetric": true,71          "type": "float"72        }73      }74    },75    "format": "float-quantized",76    "global_compression_ratio": 1.950531921411554,77    "ignore": [],78    "kv_cache_scheme": null,79    "quant_method": "compressed-tensors",80    "quantization_status": "compressed"81  },82  "scale_embedding": false,83  "torch_dtype": "float32",84  "transformers_version": "4.49.0",85  "use_cache": true,86  "use_weighted_layer_sum": false,87  "vocab_size": 5186588}