RedHatAI/diffusiongemma-26B-A4B-it-FP8-dynamic
1247k
1{2 "architectures": [3 "DiffusionGemmaForBlockDiffusion"4 ],5 "boi_token_id": 255999,6 "canvas_length": 256,7 "dtype": "bfloat16",8 "eoi_token_id": 258882,9 "eos_token_id": [10 1,11 10612 ],13 "image_token_id": 258880,14 "initializer_range": 0.02,15 "model_type": "diffusion_gemma",16 "quantization_config": {17 "config_groups": {18 "group_0": {19 "format": "float-quantized",20 "input_activations": {21 "actorder": null,22 "block_structure": null,23 "dynamic": true,24 "group_size": null,25 "num_bits": 8,26 "observer": null,27 "observer_kwargs": {},28 "scale_dtype": null,29 "strategy": "token",30 "symmetric": true,31 "type": "float",32 "zp_dtype": null33 },34 "output_activations": null,35 "targets": [36 "Linear"37 ],38 "weights": {39 "actorder": null,40 "block_structure": null,41 "dynamic": false,42 "group_size": null,43 "num_bits": 8,44 "observer": "memoryless_minmax",45 "observer_kwargs": {},46 "scale_dtype": null,47 "strategy": "channel",48 "symmetric": true,49 "type": "float",50 "zp_dtype": null51 }52 }53 },54 "format": "float-quantized",55 "global_compression_ratio": null,56 "ignore": [57 "model.encoder.language_model.layers.0.router.proj",58 "model.encoder.language_model.layers.1.router.proj",59 "model.encoder.language_model.layers.2.router.proj",60 "model.encoder.language_model.layers.3.router.proj",61 "model.encoder.language_model.layers.4.router.proj",62 "model.encoder.language_model.layers.5.router.proj",63 "model.encoder.language_model.layers.6.router.proj",64 "model.encoder.language_model.layers.7.router.proj",65 "model.encoder.language_model.layers.8.router.proj",66 "model.encoder.language_model.layers.9.router.proj",67 "model.encoder.language_model.layers.10.router.proj",68 "model.encoder.language_model.layers.11.router.proj",69 "model.encoder.language_model.layers.12.router.proj",70 "model.encoder.language_model.layers.13.router.proj",71 "model.encoder.language_model.layers.14.router.proj",72 "model.encoder.language_model.layers.15.router.proj",73 "model.encoder.language_model.layers.16.router.proj",74 "model.encoder.language_model.layers.17.router.proj",75 "model.encoder.language_model.layers.18.router.proj",76 "model.encoder.language_model.layers.19.router.proj",77 "model.encoder.language_model.layers.20.router.proj",78 "model.encoder.language_model.layers.21.router.proj",79 "model.encoder.language_model.layers.22.router.proj",80 "model.encoder.language_model.layers.23.router.proj",81 "model.encoder.language_model.layers.24.router.proj",82 "model.encoder.language_model.layers.25.router.proj",83 "model.encoder.language_model.layers.26.router.proj",84 "model.encoder.language_model.layers.27.router.proj",85 "model.encoder.language_model.layers.28.router.proj",86 "model.encoder.language_model.layers.29.router.proj",87 "model.encoder.vision_tower.patch_embedder.input_proj",88 "model.encoder.vision_tower.encoder.layers.0.self_attn.q_proj.linear",89 "model.encoder.vision_tower.encoder.layers.0.self_attn.k_proj.linear",90 "model.encoder.vision_tower.encoder.layers.0.self_attn.v_proj.linear",91 "model.encoder.vision_tower.encoder.layers.0.self_attn.o_proj.linear",92 "model.encoder.vision_tower.encoder.layers.0.mlp.gate_proj.linear",93 "model.encoder.vision_tower.encoder.layers.0.mlp.up_proj.linear",94 "model.encoder.vision_tower.encoder.layers.0.mlp.down_proj.linear",95 "model.encoder.vision_tower.encoder.layers.1.self_attn.q_proj.linear",96 "model.encoder.vision_tower.encoder.layers.1.self_attn.k_proj.linear",97 "model.encoder.vision_tower.encoder.layers.1.self_attn.v_proj.linear",98 "model.encoder.vision_tower.encoder.layers.1.self_attn.o_proj.linear",99 "model.encoder.vision_tower.encoder.layers.1.mlp.gate_proj.linear",100 "model.encoder.vision_tower.encoder.layers.1.mlp.up_proj.linear",101 "model.encoder.vision_tower.encoder.layers.1.mlp.down_proj.linear",102 "model.encoder.vision_tower.encoder.layers.2.self_attn.q_proj.linear",103 "model.encoder.vision_tower.encoder.layers.2.self_attn.k_proj.linear",104 "model.encoder.vision_tower.encoder.layers.2.self_attn.v_proj.linear",105 "model.encoder.vision_tower.encoder.layers.2.self_attn.o_proj.linear",106 "model.encoder.vision_tower.encoder.layers.2.mlp.gate_proj.linear",107 "model.encoder.vision_tower.encoder.layers.2.mlp.up_proj.linear",108 "model.encoder.vision_tower.encoder.layers.2.mlp.down_proj.linear",109 "model.encoder.vision_tower.encoder.layers.3.self_attn.q_proj.linear",110 "model.encoder.vision_tower.encoder.layers.3.self_attn.k_proj.linear",111 "model.encoder.vision_tower.encoder.layers.3.self_attn.v_proj.linear",112 "model.encoder.vision_tower.encoder.layers.3.self_attn.o_proj.linear",113 "model.encoder.vision_tower.encoder.layers.3.mlp.gate_proj.linear",114 "model.encoder.vision_tower.encoder.layers.3.mlp.up_proj.linear",115 "model.encoder.vision_tower.encoder.layers.3.mlp.down_proj.linear",116 "model.encoder.vision_tower.encoder.layers.4.self_attn.q_proj.linear",117 "model.encoder.vision_tower.encoder.layers.4.self_attn.k_proj.linear",118 "model.encoder.vision_tower.encoder.layers.4.self_attn.v_proj.linear",119 "model.encoder.vision_tower.encoder.layers.4.self_attn.o_proj.linear",120 "model.encoder.vision_tower.encoder.layers.4.mlp.gate_proj.linear",121 "model.encoder.vision_tower.encoder.layers.4.mlp.up_proj.linear",122 "model.encoder.vision_tower.encoder.layers.4.mlp.down_proj.linear",123 "model.encoder.vision_tower.encoder.layers.5.self_attn.q_proj.linear",124 "model.encoder.vision_tower.encoder.layers.5.self_attn.k_proj.linear",125 "model.encoder.vision_tower.encoder.layers.5.self_attn.v_proj.linear",126 "model.encoder.vision_tower.encoder.layers.5.self_attn.o_proj.linear",127 "model.encoder.vision_tower.encoder.layers.5.mlp.gate_proj.linear",128 "model.encoder.vision_tower.encoder.layers.5.mlp.up_proj.linear",129 "model.encoder.vision_tower.encoder.layers.5.mlp.down_proj.linear",130 "model.encoder.vision_tower.encoder.layers.6.self_attn.q_proj.linear",131 "model.encoder.vision_tower.encoder.layers.6.self_attn.k_proj.linear",132 "model.encoder.vision_tower.encoder.layers.6.self_attn.v_proj.linear",133 "model.encoder.vision_tower.encoder.layers.6.self_attn.o_proj.linear",134 "model.encoder.vision_tower.encoder.layers.6.mlp.gate_proj.linear",135 "model.encoder.vision_tower.encoder.layers.6.mlp.up_proj.linear",136 "model.encoder.vision_tower.encoder.layers.6.mlp.down_proj.linear",137 "model.encoder.vision_tower.encoder.layers.7.self_attn.q_proj.linear",138 "model.encoder.vision_tower.encoder.layers.7.self_attn.k_proj.linear",139 "model.encoder.vision_tower.encoder.layers.7.self_attn.v_proj.linear",140 "model.encoder.vision_tower.encoder.layers.7.self_attn.o_proj.linear",141 "model.encoder.vision_tower.encoder.layers.7.mlp.gate_proj.linear",142 "model.encoder.vision_tower.encoder.layers.7.mlp.up_proj.linear",143 "model.encoder.vision_tower.encoder.layers.7.mlp.down_proj.linear",144 "model.encoder.vision_tower.encoder.layers.8.self_attn.q_proj.linear",145 "model.encoder.vision_tower.encoder.layers.8.self_attn.k_proj.linear",146 "model.encoder.vision_tower.encoder.layers.8.self_attn.v_proj.linear",147 "model.encoder.vision_tower.encoder.layers.8.self_attn.o_proj.linear",148 "model.encoder.vision_tower.encoder.layers.8.mlp.gate_proj.linear",149 "model.encoder.vision_tower.encoder.layers.8.mlp.up_proj.linear",150 "model.encoder.vision_tower.encoder.layers.8.mlp.down_proj.linear",151 "model.encoder.vision_tower.encoder.layers.9.self_attn.q_proj.linear",152 "model.encoder.vision_tower.encoder.layers.9.self_attn.k_proj.linear",153 "model.encoder.vision_tower.encoder.layers.9.self_attn.v_proj.linear",154 "model.encoder.vision_tower.encoder.layers.9.self_attn.o_proj.linear",155 "model.encoder.vision_tower.encoder.layers.9.mlp.gate_proj.linear",156 "model.encoder.vision_tower.encoder.layers.9.mlp.up_proj.linear",157 "model.encoder.vision_tower.encoder.layers.9.mlp.down_proj.linear",158 "model.encoder.vision_tower.encoder.layers.10.self_attn.q_proj.linear",159 "model.encoder.vision_tower.encoder.layers.10.self_attn.k_proj.linear",160 "model.encoder.vision_tower.encoder.layers.10.self_attn.v_proj.linear",161 "model.encoder.vision_tower.encoder.layers.10.self_attn.o_proj.linear",162 "model.encoder.vision_tower.encoder.layers.10.mlp.gate_proj.linear",163 "model.encoder.vision_tower.encoder.layers.10.mlp.up_proj.linear",164 "model.encoder.vision_tower.encoder.layers.10.mlp.down_proj.linear",165 "model.encoder.vision_tower.encoder.layers.11.self_attn.q_proj.linear",166 "model.encoder.vision_tower.encoder.layers.11.self_attn.k_proj.linear",167 "model.encoder.vision_tower.encoder.layers.11.self_attn.v_proj.linear",168 "model.encoder.vision_tower.encoder.layers.11.self_attn.o_proj.linear",169 "model.encoder.vision_tower.encoder.layers.11.mlp.gate_proj.linear",170 "model.encoder.vision_tower.encoder.layers.11.mlp.up_proj.linear",171 "model.encoder.vision_tower.encoder.layers.11.mlp.down_proj.linear",172 "model.encoder.vision_tower.encoder.layers.12.self_attn.q_proj.linear",173 "model.encoder.vision_tower.encoder.layers.12.self_attn.k_proj.linear",174 "model.encoder.vision_tower.encoder.layers.12.self_attn.v_proj.linear",175 "model.encoder.vision_tower.encoder.layers.12.self_attn.o_proj.linear",176 "model.encoder.vision_tower.encoder.layers.12.mlp.gate_proj.linear",177 "model.encoder.vision_tower.encoder.layers.12.mlp.up_proj.linear",178 "model.encoder.vision_tower.encoder.layers.12.mlp.down_proj.linear",179 "model.encoder.vision_tower.encoder.layers.13.self_attn.q_proj.linear",180 "model.encoder.vision_tower.encoder.layers.13.self_attn.k_proj.linear",181 "model.encoder.vision_tower.encoder.layers.13.self_attn.v_proj.linear",182 "model.encoder.vision_tower.encoder.layers.13.self_attn.o_proj.linear",183 "model.encoder.vision_tower.encoder.layers.13.mlp.gate_proj.linear",184 "model.encoder.vision_tower.encoder.layers.13.mlp.up_proj.linear",185 "model.encoder.vision_tower.encoder.layers.13.mlp.down_proj.linear",186 "model.encoder.vision_tower.encoder.layers.14.self_attn.q_proj.linear",187 "model.encoder.vision_tower.encoder.layers.14.self_attn.k_proj.linear",188 "model.encoder.vision_tower.encoder.layers.14.self_attn.v_proj.linear",189 "model.encoder.vision_tower.encoder.layers.14.self_attn.o_proj.linear",190 "model.encoder.vision_tower.encoder.layers.14.mlp.gate_proj.linear",191 "model.encoder.vision_tower.encoder.layers.14.mlp.up_proj.linear",192 "model.encoder.vision_tower.encoder.layers.14.mlp.down_proj.linear",193 "model.encoder.vision_tower.encoder.layers.15.self_attn.q_proj.linear",194 "model.encoder.vision_tower.encoder.layers.15.self_attn.k_proj.linear",195 "model.encoder.vision_tower.encoder.layers.15.self_attn.v_proj.linear",196 "model.encoder.vision_tower.encoder.layers.15.self_attn.o_proj.linear",197 "model.encoder.vision_tower.encoder.layers.15.mlp.gate_proj.linear",198 "model.encoder.vision_tower.encoder.layers.15.mlp.up_proj.linear",199 "model.encoder.vision_tower.encoder.layers.15.mlp.down_proj.linear",200 "model.encoder.vision_tower.encoder.layers.16.self_attn.q_proj.linear",201 "model.encoder.vision_tower.encoder.layers.16.self_attn.k_proj.linear",202 "model.encoder.vision_tower.encoder.layers.16.self_attn.v_proj.linear",203 "model.encoder.vision_tower.encoder.layers.16.self_attn.o_proj.linear",204 "model.encoder.vision_tower.encoder.layers.16.mlp.gate_proj.linear",205 "model.encoder.vision_tower.encoder.layers.16.mlp.up_proj.linear",206 "model.encoder.vision_tower.encoder.layers.16.mlp.down_proj.linear",207 "model.encoder.vision_tower.encoder.layers.17.self_attn.q_proj.linear",208 "model.encoder.vision_tower.encoder.layers.17.self_attn.k_proj.linear",209 "model.encoder.vision_tower.encoder.layers.17.self_attn.v_proj.linear",210 "model.encoder.vision_tower.encoder.layers.17.self_attn.o_proj.linear",211 "model.encoder.vision_tower.encoder.layers.17.mlp.gate_proj.linear",212 "model.encoder.vision_tower.encoder.layers.17.mlp.up_proj.linear",213 "model.encoder.vision_tower.encoder.layers.17.mlp.down_proj.linear",214 "model.encoder.vision_tower.encoder.layers.18.self_attn.q_proj.linear",215 "model.encoder.vision_tower.encoder.layers.18.self_attn.k_proj.linear",216 "model.encoder.vision_tower.encoder.layers.18.self_attn.v_proj.linear",217 "model.encoder.vision_tower.encoder.layers.18.self_attn.o_proj.linear",218 "model.encoder.vision_tower.encoder.layers.18.mlp.gate_proj.linear",219 "model.encoder.vision_tower.encoder.layers.18.mlp.up_proj.linear",220 "model.encoder.vision_tower.encoder.layers.18.mlp.down_proj.linear",221 "model.encoder.vision_tower.encoder.layers.19.self_attn.q_proj.linear",222 "model.encoder.vision_tower.encoder.layers.19.self_attn.k_proj.linear",223 "model.encoder.vision_tower.encoder.layers.19.self_attn.v_proj.linear",224 "model.encoder.vision_tower.encoder.layers.19.self_attn.o_proj.linear",225 "model.encoder.vision_tower.encoder.layers.19.mlp.gate_proj.linear",226 "model.encoder.vision_tower.encoder.layers.19.mlp.up_proj.linear",227 "model.encoder.vision_tower.encoder.layers.19.mlp.down_proj.linear",228 "model.encoder.vision_tower.encoder.layers.20.self_attn.q_proj.linear",229 "model.encoder.vision_tower.encoder.layers.20.self_attn.k_proj.linear",230 "model.encoder.vision_tower.encoder.layers.20.self_attn.v_proj.linear",231 "model.encoder.vision_tower.encoder.layers.20.self_attn.o_proj.linear",232 "model.encoder.vision_tower.encoder.layers.20.mlp.gate_proj.linear",233 "model.encoder.vision_tower.encoder.layers.20.mlp.up_proj.linear",234 "model.encoder.vision_tower.encoder.layers.20.mlp.down_proj.linear",235 "model.encoder.vision_tower.encoder.layers.21.self_attn.q_proj.linear",236 "model.encoder.vision_tower.encoder.layers.21.self_attn.k_proj.linear",237 "model.encoder.vision_tower.encoder.layers.21.self_attn.v_proj.linear",238 "model.encoder.vision_tower.encoder.layers.21.self_attn.o_proj.linear",239 "model.encoder.vision_tower.encoder.layers.21.mlp.gate_proj.linear",240 "model.encoder.vision_tower.encoder.layers.21.mlp.up_proj.linear",241 "model.encoder.vision_tower.encoder.layers.21.mlp.down_proj.linear",242 "model.encoder.vision_tower.encoder.layers.22.self_attn.q_proj.linear",243 "model.encoder.vision_tower.encoder.layers.22.self_attn.k_proj.linear",244 "model.encoder.vision_tower.encoder.layers.22.self_attn.v_proj.linear",245 "model.encoder.vision_tower.encoder.layers.22.self_attn.o_proj.linear",246 "model.encoder.vision_tower.encoder.layers.22.mlp.gate_proj.linear",247 "model.encoder.vision_tower.encoder.layers.22.mlp.up_proj.linear",248 "model.encoder.vision_tower.encoder.layers.22.mlp.down_proj.linear",249 "model.encoder.vision_tower.encoder.layers.23.self_attn.q_proj.linear",250 "model.encoder.vision_tower.encoder.layers.23.self_attn.k_proj.linear",251 "model.encoder.vision_tower.encoder.layers.23.self_attn.v_proj.linear",252 "model.encoder.vision_tower.encoder.layers.23.self_attn.o_proj.linear",253 "model.encoder.vision_tower.encoder.layers.23.mlp.gate_proj.linear",254 "model.encoder.vision_tower.encoder.layers.23.mlp.up_proj.linear",255 "model.encoder.vision_tower.encoder.layers.23.mlp.down_proj.linear",256 "model.encoder.vision_tower.encoder.layers.24.self_attn.q_proj.linear",257 "model.encoder.vision_tower.encoder.layers.24.self_attn.k_proj.linear",258 "model.encoder.vision_tower.encoder.layers.24.self_attn.v_proj.linear",259 "model.encoder.vision_tower.encoder.layers.24.self_attn.o_proj.linear",260 "model.encoder.vision_tower.encoder.layers.24.mlp.gate_proj.linear",261 "model.encoder.vision_tower.encoder.layers.24.mlp.up_proj.linear",262 "model.encoder.vision_tower.encoder.layers.24.mlp.down_proj.linear",263 "model.encoder.vision_tower.encoder.layers.25.self_attn.q_proj.linear",264 "model.encoder.vision_tower.encoder.layers.25.self_attn.k_proj.linear",265 "model.encoder.vision_tower.encoder.layers.25.self_attn.v_proj.linear",266 "model.encoder.vision_tower.encoder.layers.25.self_attn.o_proj.linear",267 "model.encoder.vision_tower.encoder.layers.25.mlp.gate_proj.linear",268 "model.encoder.vision_tower.encoder.layers.25.mlp.up_proj.linear",269 "model.encoder.vision_tower.encoder.layers.25.mlp.down_proj.linear",270 "model.encoder.vision_tower.encoder.layers.26.self_attn.q_proj.linear",271 "model.encoder.vision_tower.encoder.layers.26.self_attn.k_proj.linear",272 "model.encoder.vision_tower.encoder.layers.26.self_attn.v_proj.linear",273 "model.encoder.vision_tower.encoder.layers.26.self_attn.o_proj.linear",274 "model.encoder.vision_tower.encoder.layers.26.mlp.gate_proj.linear",275 "model.encoder.vision_tower.encoder.layers.26.mlp.up_proj.linear",276 "model.encoder.vision_tower.encoder.layers.26.mlp.down_proj.linear",277 "model.encoder.embed_vision.embedding_projection",278 "model.decoder.layers.0.router.proj",279 "model.decoder.layers.1.router.proj",280 "model.decoder.layers.2.router.proj",281 "model.decoder.layers.3.router.proj",282 "model.decoder.layers.4.router.proj",283 "model.decoder.layers.5.router.proj",284 "model.decoder.layers.6.router.proj",285 "model.decoder.layers.7.router.proj",286 "model.decoder.layers.8.router.proj",287 "model.decoder.layers.9.router.proj",288 "model.decoder.layers.10.router.proj",289 "model.decoder.layers.11.router.proj",290 "model.decoder.layers.12.router.proj",291 "model.decoder.layers.13.router.proj",292 "model.decoder.layers.14.router.proj",293 "model.decoder.layers.15.router.proj",294 "model.decoder.layers.16.router.proj",295 "model.decoder.layers.17.router.proj",296 "model.decoder.layers.18.router.proj",297 "model.decoder.layers.19.router.proj",298 "model.decoder.layers.20.router.proj",299 "model.decoder.layers.21.router.proj",300 "model.decoder.layers.22.router.proj",301 "model.decoder.layers.23.router.proj",302 "model.decoder.layers.24.router.proj",303 "model.decoder.layers.25.router.proj",304 "model.decoder.layers.26.router.proj",305 "model.decoder.layers.27.router.proj",306 "model.decoder.layers.28.router.proj",307 "model.decoder.layers.29.router.proj",308 "model.decoder.self_conditioning.gate_proj",309 "model.decoder.self_conditioning.up_proj",310 "model.decoder.self_conditioning.down_proj",311 "lm_head"312 ],313 "kv_cache_scheme": null,314 "quant_method": "compressed-tensors",315 "quantization_status": "compressed",316 "sparsity_config": {},317 "transform_config": {},318 "version": "0.15.1.dev25+gc170de6"319 },320 "text_config": {321 "attention_bias": false,322 "attention_dropout": 0.0,323 "bos_token_id": 2,324 "dtype": "bfloat16",325 "eos_token_id": 1,326 "final_logit_softcapping": 30.0,327 "global_head_dim": 512,328 "head_dim": 256,329 "hidden_activation": "gelu_pytorch_tanh",330 "hidden_size": 2816,331 "initializer_range": 0.02,332 "intermediate_size": 2112,333 "layer_types": [334 "sliding_attention",335 "sliding_attention",336 "sliding_attention",337 "sliding_attention",338 "sliding_attention",339 "full_attention",340 "sliding_attention",341 "sliding_attention",342 "sliding_attention",343 "sliding_attention",344 "sliding_attention",345 "full_attention",346 "sliding_attention",347 "sliding_attention",348 "sliding_attention",349 "sliding_attention",350 "sliding_attention",351 "full_attention",352 "sliding_attention",353 "sliding_attention",354 "sliding_attention",355 "sliding_attention",356 "sliding_attention",357 "full_attention",358 "sliding_attention",359 "sliding_attention",360 "sliding_attention",361 "sliding_attention",362 "sliding_attention",363 "full_attention"364 ],365 "max_position_embeddings": 262144,366 "model_type": "diffusion_gemma_text",367 "moe_intermediate_size": 704,368 "num_attention_heads": 16,369 "num_experts": 128,370 "num_global_key_value_heads": 2,371 "num_hidden_layers": 30,372 "num_key_value_heads": 8,373 "pad_token_id": 0,374 "rms_norm_eps": 1e-06,375 "rope_parameters": {376 "full_attention": {377 "partial_rotary_factor": 0.25,378 "rope_theta": 1000000.0,379 "rope_type": "proportional"380 },381 "sliding_attention": {382 "rope_theta": 10000.0,383 "rope_type": "default"384 }385 },386 "sliding_window": 1024,387 "tie_word_embeddings": true,388 "top_k_experts": 8,389 "use_bidirectional_attention": "vision",390 "vocab_size": 262144391 },392 "tie_word_embeddings": true,393 "transformers_version": "5.8.0.dev0",394 "vision_config": {395 "_name_or_path": "",396 "architectures": null,397 "attention_bias": false,398 "attention_dropout": 0.0,399 "chunk_size_feed_forward": 0,400 "default_output_length": 280,401 "dtype": "bfloat16",402 "global_head_dim": 72,403 "head_dim": 72,404 "hidden_activation": "gelu_pytorch_tanh",405 "hidden_size": 1152,406 "id2label": {407 "0": "LABEL_0",408 "1": "LABEL_1"409 },410 "initializer_range": 0.02,411 "intermediate_size": 4304,412 "is_encoder_decoder": false,413 "label2id": {414 "LABEL_0": 0,415 "LABEL_1": 1416 },417 "max_position_embeddings": 131072,418 "model_type": "gemma4_vision",419 "num_attention_heads": 16,420 "num_hidden_layers": 27,421 "num_key_value_heads": 16,422 "output_attentions": false,423 "output_hidden_states": false,424 "patch_size": 16,425 "pooling_kernel_size": 3,426 "position_embedding_size": 10240,427 "problem_type": null,428 "return_dict": true,429 "rms_norm_eps": 1e-06,430 "rope_parameters": {431 "rope_theta": 100.0,432 "rope_type": "default"433 },434 "standardize": true,435 "use_clipped_linears": false436 },437 "vision_soft_tokens_per_image": 280438}