InvokeAI/ip_adapter_sdxl_image_encoder
205.1k
1{2 "architectures": [3 "CLIPVisionModelWithProjection"4 ],5 "_name_or_path": "",6 "add_cross_attention": false,7 "attention_dropout": 0.0,8 "bad_words_ids": null,9 "begin_suppress_tokens": null,10 "bos_token_id": null,11 "chunk_size_feed_forward": 0,12 "cross_attention_hidden_size": null,13 "decoder_start_token_id": null,14 "diversity_penalty": 0.0,15 "do_sample": false,16 "dropout": 0.0,17 "early_stopping": false,18 "encoder_no_repeat_ngram_size": 0,19 "eos_token_id": null,20 "exponential_decay_length_penalty": null,21 "finetuning_task": null,22 "forced_bos_token_id": null,23 "forced_eos_token_id": null,24 "hidden_act": "gelu",25 "hidden_size": 1664,26 "id2label": {27 "0": "LABEL_0",28 "1": "LABEL_1"29 },30 "image_size": 224,31 "initializer_factor": 1.0,32 "initializer_range": 0.02,33 "intermediate_size": 8192,34 "is_decoder": false,35 "is_encoder_decoder": false,36 "label2id": {37 "LABEL_0": 0,38 "LABEL_1": 139 },40 "layer_norm_eps": 1e-05,41 "length_penalty": 1.0,42 "max_length": 20,43 "min_length": 0,44 "model_type": "clip_vision_model",45 "no_repeat_ngram_size": 0,46 "num_attention_heads": 16,47 "num_beam_groups": 1,48 "num_beams": 1,49 "num_channels": 3,50 "num_hidden_layers": 48,51 "num_return_sequences": 1,52 "output_attentions": false,53 "output_hidden_states": false,54 "output_scores": false,55 "pad_token_id": null,56 "patch_size": 14,57 "prefix": null,58 "problem_type": null,59 "pruned_heads": {},60 "remove_invalid_values": false,61 "repetition_penalty": 1.0,62 "return_dict": true,63 "return_dict_in_generate": false,64 "sep_token_id": null,65 "suppress_tokens": null,66 "task_specific_params": null,67 "temperature": 1.0,68 "tf_legacy_loss": false,69 "tie_encoder_decoder": false,70 "tie_word_embeddings": true,71 "tokenizer_class": null,72 "top_k": 50,73 "top_p": 1.0,74 "torch_dtype": null,75 "torchscript": false,76 "transformers_version": "4.24.0",77 "typical_p": 1.0,78 "use_bfloat16": false,79 "projection_dim": 128080}81 