CoolFace
Modelpublic

Echo-Team/tiny-echo-wm-flash-diffusers

sourceHugging Faceupdated 18d agoView on Hugging Face
0likes43downloads
config.json79 linesDownload Raw Back to text_encoder
1{2  "architectures": [3    "Gemma3ForConditionalGeneration"4  ],5  "boi_token_index": 255999,6  "dtype": "float32",7  "eoi_token_index": 256000,8  "eos_token_id": [9    1,10    10611  ],12  "image_token_index": 262144,13  "initializer_range": 0.02,14  "mm_tokens_per_image": 256,15  "model_type": "gemma3",16  "text_config": {17    "_sliding_window_pattern": 2,18    "attention_bias": false,19    "attention_dropout": 0.0,20    "attn_logit_softcapping": null,21    "bos_token_id": 2,22    "cache_implementation": "hybrid",23    "dtype": "float32",24    "eos_token_id": 1,25    "final_logit_softcapping": null,26    "head_dim": 32,27    "hidden_activation": "gelu_pytorch_tanh",28    "hidden_size": 32,29    "initializer_range": 0.02,30    "intermediate_size": 128,31    "layer_types": [32      "sliding_attention",33      "full_attention"34    ],35    "max_position_embeddings": 131072,36    "model_type": "gemma3_text",37    "num_attention_heads": 1,38    "num_hidden_layers": 2,39    "num_key_value_heads": 1,40    "pad_token_id": 0,41    "query_pre_attn_scalar": 168,42    "rms_norm_eps": 1e-06,43    "rope_parameters": {44      "full_attention": {45        "factor": 8.0,46        "rope_theta": 1000000.0,47        "rope_type": "linear"48      },49      "sliding_attention": {50        "rope_theta": 10000.0,51        "rope_type": "default"52      }53    },54    "sliding_window": 1024,55    "sliding_window_pattern": 2,56    "tie_word_embeddings": true,57    "use_bidirectional_attention": false,58    "use_cache": true,59    "vocab_size": 26220860  },61  "tie_word_embeddings": true,62  "transformers_version": "5.16.1",63  "vision_config": {64    "attention_dropout": 0.0,65    "dtype": "float32",66    "hidden_act": "gelu_pytorch_tanh",67    "hidden_size": 32,68    "image_size": 896,69    "intermediate_size": 128,70    "layer_norm_eps": 1e-06,71    "model_type": "siglip_vision_model",72    "num_attention_heads": 1,73    "num_channels": 3,74    "num_hidden_layers": 2,75    "patch_size": 14,76    "vision_use_head": false77  }78}79