KaLM-Embedding/KaLM-Reranker-V1-Nano-R2
292
1{2 "architectures": [3 "T5Gemma2ForConditionalGeneration"4 ],5 "bos_token_id": 2,6 "classifier_dropout_rate": 0.0,7 "decoder": {8 "_sliding_window_pattern": 6,9 "attention_bias": false,10 "attention_dropout": 0.0,11 "attn_logit_softcapping": null,12 "bos_token_id": 2,13 "dropout_rate": 0.0,14 "dtype": "bfloat16",15 "eos_token_id": 1,16 "final_logit_softcapping": null,17 "head_dim": 256,18 "hidden_activation": "gelu_pytorch_tanh",19 "hidden_size": 640,20 "initializer_range": 0.02,21 "intermediate_size": 2048,22 "layer_types": [23 "sliding_attention",24 "sliding_attention",25 "sliding_attention",26 "sliding_attention",27 "sliding_attention",28 "full_attention",29 "sliding_attention",30 "sliding_attention",31 "sliding_attention",32 "sliding_attention",33 "sliding_attention",34 "full_attention",35 "sliding_attention",36 "sliding_attention",37 "sliding_attention",38 "sliding_attention",39 "sliding_attention",40 "full_attention"41 ],42 "max_position_embeddings": 32768,43 "model_type": "t5gemma2_decoder",44 "num_attention_heads": 4,45 "num_hidden_layers": 18,46 "num_key_value_heads": 1,47 "pad_token_id": 0,48 "query_pre_attn_scalar": 256,49 "rms_norm_eps": 1e-06,50 "rope_parameters": {51 "full_attention": {52 "factor": 8.0,53 "rope_theta": 1000000,54 "rope_type": "linear"55 },56 "sliding_attention": {57 "rope_theta": 10000,58 "rope_type": "default"59 }60 },61 "sliding_window": 512,62 "use_bidirectional_attention": false,63 "use_cache": true,64 "vocab_size": 26214465 },66 "dtype": "bfloat16",67 "encoder": {68 "attention_dropout": 0.0,69 "boi_token_index": 255999,70 "dropout_rate": 0.0,71 "dtype": "bfloat16",72 "eoi_token_index": 256000,73 "image_token_index": 256001,74 "initializer_range": 0.02,75 "mm_tokens_per_image": 256,76 "model_type": "t5gemma2_encoder",77 "text_config": {78 "_name_or_path": "",79 "_sliding_window_pattern": 6,80 "add_cross_attention": false,81 "architectures": null,82 "attention_bias": false,83 "attention_dropout": 0.0,84 "attn_logit_softcapping": null,85 "bos_token_id": 2,86 "chunk_size_feed_forward": 0,87 "cross_attention_hidden_size": null,88 "decoder_start_token_id": null,89 "dropout_rate": 0.0,90 "dtype": "bfloat16",91 "eos_token_id": 1,92 "final_logit_softcapping": null,93 "finetuning_task": null,94 "head_dim": 256,95 "hidden_activation": "gelu_pytorch_tanh",96 "hidden_size": 640,97 "id2label": {98 "0": "LABEL_0",99 "1": "LABEL_1"100 },101 "initializer_range": 0.02,102 "intermediate_size": 2048,103 "is_decoder": false,104 "is_encoder_decoder": false,105 "label2id": {106 "LABEL_0": 0,107 "LABEL_1": 1108 },109 "layer_types": [110 "sliding_attention",111 "sliding_attention",112 "sliding_attention",113 "sliding_attention",114 "sliding_attention",115 "full_attention",116 "sliding_attention",117 "sliding_attention",118 "sliding_attention",119 "sliding_attention",120 "sliding_attention",121 "full_attention",122 "sliding_attention",123 "sliding_attention",124 "sliding_attention",125 "sliding_attention",126 "sliding_attention",127 "full_attention"128 ],129 "max_position_embeddings": 32768,130 "model_type": "t5gemma2_text",131 "num_attention_heads": 4,132 "num_hidden_layers": 18,133 "num_key_value_heads": 1,134 "output_attentions": false,135 "output_hidden_states": false,136 "pad_token_id": 0,137 "prefix": null,138 "problem_type": null,139 "query_pre_attn_scalar": 256,140 "return_dict": true,141 "rms_norm_eps": 1e-06,142 "rope_parameters": {143 "full_attention": {144 "factor": 8.0,145 "rope_theta": 1000000,146 "rope_type": "linear"147 },148 "sliding_attention": {149 "rope_theta": 10000,150 "rope_type": "default"151 }152 },153 "sep_token_id": null,154 "sliding_window": 512,155 "task_specific_params": null,156 "tie_encoder_decoder": false,157 "tie_word_embeddings": true,158 "tokenizer_class": null,159 "use_bidirectional_attention": false,160 "use_cache": true,161 "vocab_size": 262144162 },163 "tie_word_embeddings": true,164 "vision_config": {165 "_name_or_path": "",166 "add_cross_attention": false,167 "architectures": null,168 "attention_dropout": 0.0,169 "bos_token_id": null,170 "chunk_size_feed_forward": 0,171 "cross_attention_hidden_size": null,172 "decoder_start_token_id": null,173 "dropout_rate": 0.0,174 "dtype": "bfloat16",175 "eos_token_id": null,176 "finetuning_task": null,177 "hidden_act": "gelu_pytorch_tanh",178 "hidden_size": 1152,179 "id2label": {180 "0": "LABEL_0",181 "1": "LABEL_1"182 },183 "image_size": 896,184 "intermediate_size": 4304,185 "is_decoder": false,186 "is_encoder_decoder": false,187 "label2id": {188 "LABEL_0": 0,189 "LABEL_1": 1190 },191 "layer_norm_eps": 1e-06,192 "model_type": "siglip_vision_model",193 "num_attention_heads": 16,194 "num_channels": 3,195 "num_hidden_layers": 27,196 "output_attentions": false,197 "output_hidden_states": false,198 "pad_token_id": null,199 "patch_size": 14,200 "prefix": null,201 "problem_type": null,202 "return_dict": true,203 "sep_token_id": null,204 "task_specific_params": null,205 "tie_encoder_decoder": false,206 "tie_word_embeddings": true,207 "tokenizer_class": null,208 "vision_use_head": false,209 "vocab_size": 262144210 },211 "vocab_size": 262144212 },213 "eoi_token_index": 256000,214 "eos_token_id": 1,215 "image_token_index": 256001,216 "initializer_range": 0.02,217 "is_encoder_decoder": true,218 "model_type": "t5gemma2",219 "pad_token_id": 0,220 "tie_word_embeddings": true,221 "transformers_version": "5.3.0",222 "vocab_size": 262144223}224 