TransWithAI/Step-Audio-R1.1-NVFP4
112
1{2 "architectures": [3 "StepAudio2ForCausalLM"4 ],5 "audio_encoder_config": {6 "adapter_stride": 2,7 "kernel_size": 3,8 "llm_dim": 5120,9 "model_type": "step_audio_2_encoder",10 "n_audio_ctx": 1500,11 "n_audio_head": 20,12 "n_audio_layer": 32,13 "n_audio_state": 1280,14 "n_codebook_size": 4096,15 "n_mels": 12816 },17 "auto_map": {18 "AutoConfig": "configuration_step_audio_2.StepAudio2Config",19 "AutoModelForCausalLM": "modeling_step_audio_2.StepAudio2ForCausalLM"20 },21 "dtype": "bfloat16",22 "max_window_layers": null,23 "model_type": "step_audio_2",24 "quantization_config": {25 "config_groups": {26 "group_0": {27 "format": "nvfp4-pack-quantized",28 "input_activations": {29 "actorder": null,30 "block_structure": null,31 "dynamic": "local",32 "group_size": 16,33 "num_bits": 4,34 "observer": "static_minmax",35 "observer_kwargs": {},36 "scale_dtype": "torch.float8_e4m3fn",37 "strategy": "tensor_group",38 "symmetric": true,39 "type": "float",40 "zp_dtype": null41 },42 "output_activations": null,43 "targets": [44 "Linear"45 ],46 "weights": {47 "actorder": null,48 "block_structure": null,49 "dynamic": false,50 "group_size": 16,51 "num_bits": 4,52 "observer": "static_minmax",53 "observer_kwargs": {},54 "scale_dtype": "torch.float8_e4m3fn",55 "strategy": "tensor_group",56 "symmetric": true,57 "type": "float",58 "zp_dtype": null59 }60 }61 },62 "format": "nvfp4-pack-quantized",63 "global_compression_ratio": null,64 "ignore": [65 "encoder.blocks.0.attn.query",66 "encoder.blocks.0.attn.key",67 "encoder.blocks.0.attn.value",68 "encoder.blocks.0.attn.out",69 "encoder.blocks.0.mlp.0",70 "encoder.blocks.0.mlp.2",71 "encoder.blocks.1.attn.query",72 "encoder.blocks.1.attn.key",73 "encoder.blocks.1.attn.value",74 "encoder.blocks.1.attn.out",75 "encoder.blocks.1.mlp.0",76 "encoder.blocks.1.mlp.2",77 "encoder.blocks.2.attn.query",78 "encoder.blocks.2.attn.key",79 "encoder.blocks.2.attn.value",80 "encoder.blocks.2.attn.out",81 "encoder.blocks.2.mlp.0",82 "encoder.blocks.2.mlp.2",83 "encoder.blocks.3.attn.query",84 "encoder.blocks.3.attn.key",85 "encoder.blocks.3.attn.value",86 "encoder.blocks.3.attn.out",87 "encoder.blocks.3.mlp.0",88 "encoder.blocks.3.mlp.2",89 "encoder.blocks.4.attn.query",90 "encoder.blocks.4.attn.key",91 "encoder.blocks.4.attn.value",92 "encoder.blocks.4.attn.out",93 "encoder.blocks.4.mlp.0",94 "encoder.blocks.4.mlp.2",95 "encoder.blocks.5.attn.query",96 "encoder.blocks.5.attn.key",97 "encoder.blocks.5.attn.value",98 "encoder.blocks.5.attn.out",99 "encoder.blocks.5.mlp.0",100 "encoder.blocks.5.mlp.2",101 "encoder.blocks.6.attn.query",102 "encoder.blocks.6.attn.key",103 "encoder.blocks.6.attn.value",104 "encoder.blocks.6.attn.out",105 "encoder.blocks.6.mlp.0",106 "encoder.blocks.6.mlp.2",107 "encoder.blocks.7.attn.query",108 "encoder.blocks.7.attn.key",109 "encoder.blocks.7.attn.value",110 "encoder.blocks.7.attn.out",111 "encoder.blocks.7.mlp.0",112 "encoder.blocks.7.mlp.2",113 "encoder.blocks.8.attn.query",114 "encoder.blocks.8.attn.key",115 "encoder.blocks.8.attn.value",116 "encoder.blocks.8.attn.out",117 "encoder.blocks.8.mlp.0",118 "encoder.blocks.8.mlp.2",119 "encoder.blocks.9.attn.query",120 "encoder.blocks.9.attn.key",121 "encoder.blocks.9.attn.value",122 "encoder.blocks.9.attn.out",123 "encoder.blocks.9.mlp.0",124 "encoder.blocks.9.mlp.2",125 "encoder.blocks.10.attn.query",126 "encoder.blocks.10.attn.key",127 "encoder.blocks.10.attn.value",128 "encoder.blocks.10.attn.out",129 "encoder.blocks.10.mlp.0",130 "encoder.blocks.10.mlp.2",131 "encoder.blocks.11.attn.query",132 "encoder.blocks.11.attn.key",133 "encoder.blocks.11.attn.value",134 "encoder.blocks.11.attn.out",135 "encoder.blocks.11.mlp.0",136 "encoder.blocks.11.mlp.2",137 "encoder.blocks.12.attn.query",138 "encoder.blocks.12.attn.key",139 "encoder.blocks.12.attn.value",140 "encoder.blocks.12.attn.out",141 "encoder.blocks.12.mlp.0",142 "encoder.blocks.12.mlp.2",143 "encoder.blocks.13.attn.query",144 "encoder.blocks.13.attn.key",145 "encoder.blocks.13.attn.value",146 "encoder.blocks.13.attn.out",147 "encoder.blocks.13.mlp.0",148 "encoder.blocks.13.mlp.2",149 "encoder.blocks.14.attn.query",150 "encoder.blocks.14.attn.key",151 "encoder.blocks.14.attn.value",152 "encoder.blocks.14.attn.out",153 "encoder.blocks.14.mlp.0",154 "encoder.blocks.14.mlp.2",155 "encoder.blocks.15.attn.query",156 "encoder.blocks.15.attn.key",157 "encoder.blocks.15.attn.value",158 "encoder.blocks.15.attn.out",159 "encoder.blocks.15.mlp.0",160 "encoder.blocks.15.mlp.2",161 "encoder.blocks.16.attn.query",162 "encoder.blocks.16.attn.key",163 "encoder.blocks.16.attn.value",164 "encoder.blocks.16.attn.out",165 "encoder.blocks.16.mlp.0",166 "encoder.blocks.16.mlp.2",167 "encoder.blocks.17.attn.query",168 "encoder.blocks.17.attn.key",169 "encoder.blocks.17.attn.value",170 "encoder.blocks.17.attn.out",171 "encoder.blocks.17.mlp.0",172 "encoder.blocks.17.mlp.2",173 "encoder.blocks.18.attn.query",174 "encoder.blocks.18.attn.key",175 "encoder.blocks.18.attn.value",176 "encoder.blocks.18.attn.out",177 "encoder.blocks.18.mlp.0",178 "encoder.blocks.18.mlp.2",179 "encoder.blocks.19.attn.query",180 "encoder.blocks.19.attn.key",181 "encoder.blocks.19.attn.value",182 "encoder.blocks.19.attn.out",183 "encoder.blocks.19.mlp.0",184 "encoder.blocks.19.mlp.2",185 "encoder.blocks.20.attn.query",186 "encoder.blocks.20.attn.key",187 "encoder.blocks.20.attn.value",188 "encoder.blocks.20.attn.out",189 "encoder.blocks.20.mlp.0",190 "encoder.blocks.20.mlp.2",191 "encoder.blocks.21.attn.query",192 "encoder.blocks.21.attn.key",193 "encoder.blocks.21.attn.value",194 "encoder.blocks.21.attn.out",195 "encoder.blocks.21.mlp.0",196 "encoder.blocks.21.mlp.2",197 "encoder.blocks.22.attn.query",198 "encoder.blocks.22.attn.key",199 "encoder.blocks.22.attn.value",200 "encoder.blocks.22.attn.out",201 "encoder.blocks.22.mlp.0",202 "encoder.blocks.22.mlp.2",203 "encoder.blocks.23.attn.query",204 "encoder.blocks.23.attn.key",205 "encoder.blocks.23.attn.value",206 "encoder.blocks.23.attn.out",207 "encoder.blocks.23.mlp.0",208 "encoder.blocks.23.mlp.2",209 "encoder.blocks.24.attn.query",210 "encoder.blocks.24.attn.key",211 "encoder.blocks.24.attn.value",212 "encoder.blocks.24.attn.out",213 "encoder.blocks.24.mlp.0",214 "encoder.blocks.24.mlp.2",215 "encoder.blocks.25.attn.query",216 "encoder.blocks.25.attn.key",217 "encoder.blocks.25.attn.value",218 "encoder.blocks.25.attn.out",219 "encoder.blocks.25.mlp.0",220 "encoder.blocks.25.mlp.2",221 "encoder.blocks.26.attn.query",222 "encoder.blocks.26.attn.key",223 "encoder.blocks.26.attn.value",224 "encoder.blocks.26.attn.out",225 "encoder.blocks.26.mlp.0",226 "encoder.blocks.26.mlp.2",227 "encoder.blocks.27.attn.query",228 "encoder.blocks.27.attn.key",229 "encoder.blocks.27.attn.value",230 "encoder.blocks.27.attn.out",231 "encoder.blocks.27.mlp.0",232 "encoder.blocks.27.mlp.2",233 "encoder.blocks.28.attn.query",234 "encoder.blocks.28.attn.key",235 "encoder.blocks.28.attn.value",236 "encoder.blocks.28.attn.out",237 "encoder.blocks.28.mlp.0",238 "encoder.blocks.28.mlp.2",239 "encoder.blocks.29.attn.query",240 "encoder.blocks.29.attn.key",241 "encoder.blocks.29.attn.value",242 "encoder.blocks.29.attn.out",243 "encoder.blocks.29.mlp.0",244 "encoder.blocks.29.mlp.2",245 "encoder.blocks.30.attn.query",246 "encoder.blocks.30.attn.key",247 "encoder.blocks.30.attn.value",248 "encoder.blocks.30.attn.out",249 "encoder.blocks.30.mlp.0",250 "encoder.blocks.30.mlp.2",251 "encoder.blocks.31.attn.query",252 "encoder.blocks.31.attn.key",253 "encoder.blocks.31.attn.value",254 "encoder.blocks.31.attn.out",255 "encoder.blocks.31.mlp.0",256 "encoder.blocks.31.mlp.2",257 "adapter.linear1",258 "adapter.linear2",259 "lm_head"260 ],261 "kv_cache_scheme": null,262 "quant_method": "compressed-tensors",263 "quantization_status": "compressed",264 "sparsity_config": {},265 "transform_config": {},266 "version": "0.13.0"267 },268 "sliding_window": 2048,269 "text_config": {270 "architectures": [271 "Qwen2ForCausalLM"272 ],273 "attention_dropout": 0.0,274 "dtype": "bfloat16",275 "hidden_act": "silu",276 "hidden_size": 5120,277 "initializer_range": 0.02,278 "intermediate_size": 27648,279 "layer_types": [280 "full_attention",281 "full_attention",282 "full_attention",283 "full_attention",284 "full_attention",285 "full_attention",286 "full_attention",287 "full_attention",288 "full_attention",289 "full_attention",290 "full_attention",291 "full_attention",292 "full_attention",293 "full_attention",294 "full_attention",295 "full_attention",296 "full_attention",297 "full_attention",298 "full_attention",299 "full_attention",300 "full_attention",301 "full_attention",302 "full_attention",303 "full_attention",304 "full_attention",305 "full_attention",306 "full_attention",307 "full_attention",308 "full_attention",309 "full_attention",310 "full_attention",311 "full_attention",312 "full_attention",313 "full_attention",314 "full_attention",315 "full_attention",316 "full_attention",317 "full_attention",318 "full_attention",319 "full_attention",320 "full_attention",321 "full_attention",322 "full_attention",323 "full_attention",324 "full_attention",325 "full_attention",326 "full_attention",327 "full_attention",328 "full_attention",329 "full_attention",330 "full_attention",331 "full_attention",332 "full_attention",333 "full_attention",334 "full_attention",335 "full_attention",336 "full_attention",337 "full_attention",338 "full_attention",339 "full_attention",340 "full_attention",341 "full_attention",342 "full_attention",343 "full_attention"344 ],345 "max_position_embeddings": 65536,346 "max_window_layers": 28,347 "model_type": "qwen2",348 "num_attention_heads": 40,349 "num_hidden_layers": 64,350 "num_attention_groups": 8,351 "num_key_value_heads": 8,352 "rms_norm_eps": 1e-05,353 "rope_scaling": null,354 "rope_theta": 1000000.0,355 "sliding_window": null,356 "use_cache": true,357 "use_sliding_window": false,358 "vocab_size": 158720359 },360 "transformers_version": "4.57.3",361 "use_sliding_window": false362}