Firworks/Step-Audio-R1-nvfp4
217
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": "minmax",35 "observer_kwargs": {},36 "strategy": "tensor_group",37 "symmetric": true,38 "type": "float"39 },40 "output_activations": null,41 "targets": [42 "Linear"43 ],44 "weights": {45 "actorder": null,46 "block_structure": null,47 "dynamic": false,48 "group_size": 16,49 "num_bits": 4,50 "observer": "minmax",51 "observer_kwargs": {},52 "strategy": "tensor_group",53 "symmetric": true,54 "type": "float"55 }56 }57 },58 "format": "nvfp4-pack-quantized",59 "global_compression_ratio": null,60 "ignore": [61 "encoder.blocks.0.attn.query",62 "encoder.blocks.0.attn.key",63 "encoder.blocks.0.attn.value",64 "encoder.blocks.0.attn.out",65 "encoder.blocks.0.mlp.0",66 "encoder.blocks.0.mlp.2",67 "encoder.blocks.1.attn.query",68 "encoder.blocks.1.attn.key",69 "encoder.blocks.1.attn.value",70 "encoder.blocks.1.attn.out",71 "encoder.blocks.1.mlp.0",72 "encoder.blocks.1.mlp.2",73 "encoder.blocks.2.attn.query",74 "encoder.blocks.2.attn.key",75 "encoder.blocks.2.attn.value",76 "encoder.blocks.2.attn.out",77 "encoder.blocks.2.mlp.0",78 "encoder.blocks.2.mlp.2",79 "encoder.blocks.3.attn.query",80 "encoder.blocks.3.attn.key",81 "encoder.blocks.3.attn.value",82 "encoder.blocks.3.attn.out",83 "encoder.blocks.3.mlp.0",84 "encoder.blocks.3.mlp.2",85 "encoder.blocks.4.attn.query",86 "encoder.blocks.4.attn.key",87 "encoder.blocks.4.attn.value",88 "encoder.blocks.4.attn.out",89 "encoder.blocks.4.mlp.0",90 "encoder.blocks.4.mlp.2",91 "encoder.blocks.5.attn.query",92 "encoder.blocks.5.attn.key",93 "encoder.blocks.5.attn.value",94 "encoder.blocks.5.attn.out",95 "encoder.blocks.5.mlp.0",96 "encoder.blocks.5.mlp.2",97 "encoder.blocks.6.attn.query",98 "encoder.blocks.6.attn.key",99 "encoder.blocks.6.attn.value",100 "encoder.blocks.6.attn.out",101 "encoder.blocks.6.mlp.0",102 "encoder.blocks.6.mlp.2",103 "encoder.blocks.7.attn.query",104 "encoder.blocks.7.attn.key",105 "encoder.blocks.7.attn.value",106 "encoder.blocks.7.attn.out",107 "encoder.blocks.7.mlp.0",108 "encoder.blocks.7.mlp.2",109 "encoder.blocks.8.attn.query",110 "encoder.blocks.8.attn.key",111 "encoder.blocks.8.attn.value",112 "encoder.blocks.8.attn.out",113 "encoder.blocks.8.mlp.0",114 "encoder.blocks.8.mlp.2",115 "encoder.blocks.9.attn.query",116 "encoder.blocks.9.attn.key",117 "encoder.blocks.9.attn.value",118 "encoder.blocks.9.attn.out",119 "encoder.blocks.9.mlp.0",120 "encoder.blocks.9.mlp.2",121 "encoder.blocks.10.attn.query",122 "encoder.blocks.10.attn.key",123 "encoder.blocks.10.attn.value",124 "encoder.blocks.10.attn.out",125 "encoder.blocks.10.mlp.0",126 "encoder.blocks.10.mlp.2",127 "encoder.blocks.11.attn.query",128 "encoder.blocks.11.attn.key",129 "encoder.blocks.11.attn.value",130 "encoder.blocks.11.attn.out",131 "encoder.blocks.11.mlp.0",132 "encoder.blocks.11.mlp.2",133 "encoder.blocks.12.attn.query",134 "encoder.blocks.12.attn.key",135 "encoder.blocks.12.attn.value",136 "encoder.blocks.12.attn.out",137 "encoder.blocks.12.mlp.0",138 "encoder.blocks.12.mlp.2",139 "encoder.blocks.13.attn.query",140 "encoder.blocks.13.attn.key",141 "encoder.blocks.13.attn.value",142 "encoder.blocks.13.attn.out",143 "encoder.blocks.13.mlp.0",144 "encoder.blocks.13.mlp.2",145 "encoder.blocks.14.attn.query",146 "encoder.blocks.14.attn.key",147 "encoder.blocks.14.attn.value",148 "encoder.blocks.14.attn.out",149 "encoder.blocks.14.mlp.0",150 "encoder.blocks.14.mlp.2",151 "encoder.blocks.15.attn.query",152 "encoder.blocks.15.attn.key",153 "encoder.blocks.15.attn.value",154 "encoder.blocks.15.attn.out",155 "encoder.blocks.15.mlp.0",156 "encoder.blocks.15.mlp.2",157 "encoder.blocks.16.attn.query",158 "encoder.blocks.16.attn.key",159 "encoder.blocks.16.attn.value",160 "encoder.blocks.16.attn.out",161 "encoder.blocks.16.mlp.0",162 "encoder.blocks.16.mlp.2",163 "encoder.blocks.17.attn.query",164 "encoder.blocks.17.attn.key",165 "encoder.blocks.17.attn.value",166 "encoder.blocks.17.attn.out",167 "encoder.blocks.17.mlp.0",168 "encoder.blocks.17.mlp.2",169 "encoder.blocks.18.attn.query",170 "encoder.blocks.18.attn.key",171 "encoder.blocks.18.attn.value",172 "encoder.blocks.18.attn.out",173 "encoder.blocks.18.mlp.0",174 "encoder.blocks.18.mlp.2",175 "encoder.blocks.19.attn.query",176 "encoder.blocks.19.attn.key",177 "encoder.blocks.19.attn.value",178 "encoder.blocks.19.attn.out",179 "encoder.blocks.19.mlp.0",180 "encoder.blocks.19.mlp.2",181 "encoder.blocks.20.attn.query",182 "encoder.blocks.20.attn.key",183 "encoder.blocks.20.attn.value",184 "encoder.blocks.20.attn.out",185 "encoder.blocks.20.mlp.0",186 "encoder.blocks.20.mlp.2",187 "encoder.blocks.21.attn.query",188 "encoder.blocks.21.attn.key",189 "encoder.blocks.21.attn.value",190 "encoder.blocks.21.attn.out",191 "encoder.blocks.21.mlp.0",192 "encoder.blocks.21.mlp.2",193 "encoder.blocks.22.attn.query",194 "encoder.blocks.22.attn.key",195 "encoder.blocks.22.attn.value",196 "encoder.blocks.22.attn.out",197 "encoder.blocks.22.mlp.0",198 "encoder.blocks.22.mlp.2",199 "encoder.blocks.23.attn.query",200 "encoder.blocks.23.attn.key",201 "encoder.blocks.23.attn.value",202 "encoder.blocks.23.attn.out",203 "encoder.blocks.23.mlp.0",204 "encoder.blocks.23.mlp.2",205 "encoder.blocks.24.attn.query",206 "encoder.blocks.24.attn.key",207 "encoder.blocks.24.attn.value",208 "encoder.blocks.24.attn.out",209 "encoder.blocks.24.mlp.0",210 "encoder.blocks.24.mlp.2",211 "encoder.blocks.25.attn.query",212 "encoder.blocks.25.attn.key",213 "encoder.blocks.25.attn.value",214 "encoder.blocks.25.attn.out",215 "encoder.blocks.25.mlp.0",216 "encoder.blocks.25.mlp.2",217 "encoder.blocks.26.attn.query",218 "encoder.blocks.26.attn.key",219 "encoder.blocks.26.attn.value",220 "encoder.blocks.26.attn.out",221 "encoder.blocks.26.mlp.0",222 "encoder.blocks.26.mlp.2",223 "encoder.blocks.27.attn.query",224 "encoder.blocks.27.attn.key",225 "encoder.blocks.27.attn.value",226 "encoder.blocks.27.attn.out",227 "encoder.blocks.27.mlp.0",228 "encoder.blocks.27.mlp.2",229 "encoder.blocks.28.attn.query",230 "encoder.blocks.28.attn.key",231 "encoder.blocks.28.attn.value",232 "encoder.blocks.28.attn.out",233 "encoder.blocks.28.mlp.0",234 "encoder.blocks.28.mlp.2",235 "encoder.blocks.29.attn.query",236 "encoder.blocks.29.attn.key",237 "encoder.blocks.29.attn.value",238 "encoder.blocks.29.attn.out",239 "encoder.blocks.29.mlp.0",240 "encoder.blocks.29.mlp.2",241 "encoder.blocks.30.attn.query",242 "encoder.blocks.30.attn.key",243 "encoder.blocks.30.attn.value",244 "encoder.blocks.30.attn.out",245 "encoder.blocks.30.mlp.0",246 "encoder.blocks.30.mlp.2",247 "encoder.blocks.31.attn.query",248 "encoder.blocks.31.attn.key",249 "encoder.blocks.31.attn.value",250 "encoder.blocks.31.attn.out",251 "encoder.blocks.31.mlp.0",252 "encoder.blocks.31.mlp.2",253 "adapter.linear1",254 "adapter.linear2",255 "lm_head"256 ],257 "kv_cache_scheme": null,258 "quant_method": "compressed-tensors",259 "quantization_status": "compressed",260 "sparsity_config": {},261 "transform_config": {},262 "version": "0.12.2"263 },264 "sliding_window": 2048,265 "text_config": {266 "architectures": [267 "Qwen2ForCausalLM"268 ],269 "attention_dropout": 0.0,270 "dtype": "bfloat16",271 "hidden_act": "silu",272 "hidden_size": 5120,273 "initializer_range": 0.02,274 "intermediate_size": 27648,275 "layer_types": [276 "full_attention",277 "full_attention",278 "full_attention",279 "full_attention",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 ],341 "max_position_embeddings": 65536,342 "max_window_layers": 28,343 "model_type": "qwen2",344 "num_attention_heads": 40,345 "num_hidden_layers": 64,346 "num_key_value_heads": 8,347 "rms_norm_eps": 1e-05,348 "rope_scaling": null,349 "rope_theta": 1000000.0,350 "sliding_window": null,351 "use_cache": true,352 "use_sliding_window": false,353 "vocab_size": 158720354 },355 "tie_word_embeddings": false,356 "transformers_version": "4.56.2",357 "use_sliding_window": false358}