import torch from transformers import T5EncoderModel, T5Config from .sd_text_encoder import SDTextEncoder from .sdxl_text_encoder import SDXLTextEncoder2, SDXLTextEncoder2StateDictConverter class SD3TextEncoder1(SDTextEncoder): def __init__(self, vocab_size=49408): super().__init__(vocab_size=vocab_size) def forward(self, input_ids, clip_skip=2): embeds = self.token_embedding(input_ids) + self.position_embeds attn_mask = self.attn_mask.to(device=embeds.device, dtype=embeds.dtype) for encoder_id, encoder in enumerate(self.encoders): embeds = encoder(embeds, attn_mask=attn_mask) if encoder_id + clip_skip == len(self.encoders): hidden_states = embeds embeds = self.final_layer_norm(embeds) pooled_embeds = embeds[torch.arange(embeds.shape[0]), input_ids.to(dtype=torch.int).argmax(dim=-1)] return pooled_embeds, hidden_states @staticmethod def state_dict_converter(): return SD3TextEncoder1StateDictConverter() class SD3TextEncoder2(SDXLTextEncoder2): def __init__(self): super().__init__() @staticmethod def state_dict_converter(): return SD3TextEncoder2StateDictConverter() class SD3TextEncoder3(T5EncoderModel): def __init__(self): config = T5Config( _name_or_path = ".", architectures = ["T5EncoderModel"], classifier_dropout = 0.0, d_ff = 10240, d_kv = 64, d_model = 4096, decoder_start_token_id = 0, dense_act_fn = "gelu_new", dropout_rate = 0.1, eos_token_id = 1, feed_forward_proj = "gated-gelu", initializer_factor = 1.0, is_encoder_decoder = True, is_gated_act = True, layer_norm_epsilon = 1e-06, model_type = "t5", num_decoder_layers = 24, num_heads = 64, num_layers = 24, output_past = True, pad_token_id = 0, relative_attention_max_distance = 128, relative_attention_num_buckets = 32, tie_word_embeddings = False, torch_dtype = torch.float16, transformers_version = "4.41.2", use_cache = True, vocab_size = 32128 ) super().__init__(config) self.eval() def forward(self, input_ids): outputs = super().forward(input_ids=input_ids) prompt_emb = outputs.last_hidden_state return prompt_emb @staticmethod def state_dict_converter(): return SD3TextEncoder3StateDictConverter() class SD3TextEncoder1StateDictConverter: def __init__(self): pass def from_diffusers(self, state_dict): rename_dict = { "text_model.embeddings.token_embedding.weight": "token_embedding.weight", "text_model.embeddings.position_embedding.weight": "position_embeds", "text_model.final_layer_norm.weight": "final_layer_norm.weight", "text_model.final_layer_norm.bias": "final_layer_norm.bias", } attn_rename_dict = { "self_attn.q_proj": "attn.to_q", "self_attn.k_proj": "attn.to_k", "self_attn.v_proj": "attn.to_v", "self_attn.out_proj": "attn.to_out", "layer_norm1": "layer_norm1", "layer_norm2": "layer_norm2", "mlp.fc1": "fc1", "mlp.fc2": "fc2", } state_dict_ = {} for name in state_dict: if name in rename_dict: param = state_dict[name] if name == "text_model.embeddings.position_embedding.weight": param = param.reshape((1, param.shape[0], param.shape[1])) state_dict_[rename_dict[name]] = param elif name.startswith("text_model.encoder.layers."): param = state_dict[name] names = name.split(".") layer_id, layer_type, tail = names[3], ".".join(names[4:-1]), names[-1] name_ = ".".join(["encoders", layer_id, attn_rename_dict[layer_type], tail]) state_dict_[name_] = param return state_dict_ def from_civitai(self, state_dict): rename_dict = { "text_encoders.clip_l.transformer.text_model.embeddings.position_embedding.weight": "position_embeds", "text_encoders.clip_l.transformer.text_model.embeddings.token_embedding.weight": "token_embedding.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.layer_norm1.bias": "encoders.0.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.layer_norm1.weight": "encoders.0.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.layer_norm2.bias": "encoders.0.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.layer_norm2.weight": "encoders.0.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.mlp.fc1.bias": "encoders.0.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.mlp.fc1.weight": "encoders.0.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.mlp.fc2.bias": "encoders.0.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.mlp.fc2.weight": "encoders.0.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.k_proj.bias": "encoders.0.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.k_proj.weight": "encoders.0.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.out_proj.bias": "encoders.0.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.out_proj.weight": "encoders.0.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.q_proj.bias": "encoders.0.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.q_proj.weight": "encoders.0.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.v_proj.bias": "encoders.0.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.0.self_attn.v_proj.weight": "encoders.0.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.layer_norm1.bias": "encoders.1.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.layer_norm1.weight": "encoders.1.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.layer_norm2.bias": "encoders.1.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.layer_norm2.weight": "encoders.1.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.mlp.fc1.bias": "encoders.1.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.mlp.fc1.weight": "encoders.1.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.mlp.fc2.bias": "encoders.1.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.mlp.fc2.weight": "encoders.1.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.k_proj.bias": "encoders.1.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.k_proj.weight": "encoders.1.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.out_proj.bias": "encoders.1.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.out_proj.weight": "encoders.1.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.q_proj.bias": "encoders.1.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.q_proj.weight": "encoders.1.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.v_proj.bias": "encoders.1.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.1.self_attn.v_proj.weight": "encoders.1.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.layer_norm1.bias": "encoders.10.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.layer_norm1.weight": "encoders.10.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.layer_norm2.bias": "encoders.10.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.layer_norm2.weight": "encoders.10.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.mlp.fc1.bias": "encoders.10.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.mlp.fc1.weight": "encoders.10.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.mlp.fc2.bias": "encoders.10.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.mlp.fc2.weight": "encoders.10.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.k_proj.bias": "encoders.10.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.k_proj.weight": "encoders.10.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.out_proj.bias": "encoders.10.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.out_proj.weight": "encoders.10.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.q_proj.bias": "encoders.10.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.q_proj.weight": "encoders.10.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.v_proj.bias": "encoders.10.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.10.self_attn.v_proj.weight": "encoders.10.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.layer_norm1.bias": "encoders.11.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.layer_norm1.weight": "encoders.11.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.layer_norm2.bias": "encoders.11.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.layer_norm2.weight": "encoders.11.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.mlp.fc1.bias": "encoders.11.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.mlp.fc1.weight": "encoders.11.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.mlp.fc2.bias": "encoders.11.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.mlp.fc2.weight": "encoders.11.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.k_proj.bias": "encoders.11.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.k_proj.weight": "encoders.11.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.out_proj.bias": "encoders.11.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.out_proj.weight": "encoders.11.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.q_proj.bias": "encoders.11.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.q_proj.weight": "encoders.11.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.v_proj.bias": "encoders.11.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.11.self_attn.v_proj.weight": "encoders.11.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.layer_norm1.bias": "encoders.2.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.layer_norm1.weight": "encoders.2.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.layer_norm2.bias": "encoders.2.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.layer_norm2.weight": "encoders.2.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.mlp.fc1.bias": "encoders.2.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.mlp.fc1.weight": "encoders.2.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.mlp.fc2.bias": "encoders.2.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.mlp.fc2.weight": "encoders.2.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.k_proj.bias": "encoders.2.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.k_proj.weight": "encoders.2.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.out_proj.bias": "encoders.2.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.out_proj.weight": "encoders.2.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.q_proj.bias": "encoders.2.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.q_proj.weight": "encoders.2.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.v_proj.bias": "encoders.2.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.2.self_attn.v_proj.weight": "encoders.2.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.layer_norm1.bias": "encoders.3.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.layer_norm1.weight": "encoders.3.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.layer_norm2.bias": "encoders.3.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.layer_norm2.weight": "encoders.3.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.mlp.fc1.bias": "encoders.3.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.mlp.fc1.weight": "encoders.3.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.mlp.fc2.bias": "encoders.3.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.mlp.fc2.weight": "encoders.3.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.k_proj.bias": "encoders.3.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.k_proj.weight": "encoders.3.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.out_proj.bias": "encoders.3.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.out_proj.weight": "encoders.3.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.q_proj.bias": "encoders.3.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.q_proj.weight": "encoders.3.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.v_proj.bias": "encoders.3.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.3.self_attn.v_proj.weight": "encoders.3.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.layer_norm1.bias": "encoders.4.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.layer_norm1.weight": "encoders.4.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.layer_norm2.bias": "encoders.4.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.layer_norm2.weight": "encoders.4.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.mlp.fc1.bias": "encoders.4.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.mlp.fc1.weight": "encoders.4.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.mlp.fc2.bias": "encoders.4.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.mlp.fc2.weight": "encoders.4.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.k_proj.bias": "encoders.4.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.k_proj.weight": "encoders.4.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.out_proj.bias": "encoders.4.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.out_proj.weight": "encoders.4.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.q_proj.bias": "encoders.4.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.q_proj.weight": "encoders.4.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.v_proj.bias": "encoders.4.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.4.self_attn.v_proj.weight": "encoders.4.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.layer_norm1.bias": "encoders.5.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.layer_norm1.weight": "encoders.5.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.layer_norm2.bias": "encoders.5.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.layer_norm2.weight": "encoders.5.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.mlp.fc1.bias": "encoders.5.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.mlp.fc1.weight": "encoders.5.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.mlp.fc2.bias": "encoders.5.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.mlp.fc2.weight": "encoders.5.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.k_proj.bias": "encoders.5.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.k_proj.weight": "encoders.5.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.out_proj.bias": "encoders.5.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.out_proj.weight": "encoders.5.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.q_proj.bias": "encoders.5.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.q_proj.weight": "encoders.5.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.v_proj.bias": "encoders.5.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.5.self_attn.v_proj.weight": "encoders.5.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.layer_norm1.bias": "encoders.6.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.layer_norm1.weight": "encoders.6.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.layer_norm2.bias": "encoders.6.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.layer_norm2.weight": "encoders.6.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.mlp.fc1.bias": "encoders.6.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.mlp.fc1.weight": "encoders.6.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.mlp.fc2.bias": "encoders.6.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.mlp.fc2.weight": "encoders.6.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.k_proj.bias": "encoders.6.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.k_proj.weight": "encoders.6.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.out_proj.bias": "encoders.6.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.out_proj.weight": "encoders.6.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.q_proj.bias": "encoders.6.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.q_proj.weight": "encoders.6.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.v_proj.bias": "encoders.6.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.6.self_attn.v_proj.weight": "encoders.6.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.layer_norm1.bias": "encoders.7.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.layer_norm1.weight": "encoders.7.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.layer_norm2.bias": "encoders.7.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.layer_norm2.weight": "encoders.7.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.mlp.fc1.bias": "encoders.7.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.mlp.fc1.weight": "encoders.7.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.mlp.fc2.bias": "encoders.7.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.mlp.fc2.weight": "encoders.7.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.k_proj.bias": "encoders.7.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.k_proj.weight": "encoders.7.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.out_proj.bias": "encoders.7.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.out_proj.weight": "encoders.7.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.q_proj.bias": "encoders.7.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.q_proj.weight": "encoders.7.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.v_proj.bias": "encoders.7.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.7.self_attn.v_proj.weight": "encoders.7.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.layer_norm1.bias": "encoders.8.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.layer_norm1.weight": "encoders.8.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.layer_norm2.bias": "encoders.8.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.layer_norm2.weight": "encoders.8.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.mlp.fc1.bias": "encoders.8.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.mlp.fc1.weight": "encoders.8.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.mlp.fc2.bias": "encoders.8.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.mlp.fc2.weight": "encoders.8.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.k_proj.bias": "encoders.8.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.k_proj.weight": "encoders.8.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.out_proj.bias": "encoders.8.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.out_proj.weight": "encoders.8.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.q_proj.bias": "encoders.8.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.q_proj.weight": "encoders.8.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.v_proj.bias": "encoders.8.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.8.self_attn.v_proj.weight": "encoders.8.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.layer_norm1.bias": "encoders.9.layer_norm1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.layer_norm1.weight": "encoders.9.layer_norm1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.layer_norm2.bias": "encoders.9.layer_norm2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.layer_norm2.weight": "encoders.9.layer_norm2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.mlp.fc1.bias": "encoders.9.fc1.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.mlp.fc1.weight": "encoders.9.fc1.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.mlp.fc2.bias": "encoders.9.fc2.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.mlp.fc2.weight": "encoders.9.fc2.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.k_proj.bias": "encoders.9.attn.to_k.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.k_proj.weight": "encoders.9.attn.to_k.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.out_proj.bias": "encoders.9.attn.to_out.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.out_proj.weight": "encoders.9.attn.to_out.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.q_proj.bias": "encoders.9.attn.to_q.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.q_proj.weight": "encoders.9.attn.to_q.weight", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.v_proj.bias": "encoders.9.attn.to_v.bias", "text_encoders.clip_l.transformer.text_model.encoder.layers.9.self_attn.v_proj.weight": "encoders.9.attn.to_v.weight", "text_encoders.clip_l.transformer.text_model.final_layer_norm.bias": "final_layer_norm.bias", "text_encoders.clip_l.transformer.text_model.final_layer_norm.weight": "final_layer_norm.weight", } state_dict_ = {} for name in state_dict: if name in rename_dict: param = state_dict[name] if name == "text_encoders.clip_l.transformer.text_model.embeddings.position_embedding.weight": param = param.reshape((1, param.shape[0], param.shape[1])) state_dict_[rename_dict[name]] = param return state_dict_ class SD3TextEncoder2StateDictConverter(SDXLTextEncoder2StateDictConverter): def __init__(self): pass def from_diffusers(self, state_dict): return super().from_diffusers(state_dict) def from_civitai(self, state_dict): rename_dict = { "text_encoders.clip_g.transformer.text_model.embeddings.position_embedding.weight": "position_embeds", "text_encoders.clip_g.transformer.text_model.embeddings.token_embedding.weight": "token_embedding.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.layer_norm1.bias": "encoders.0.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.layer_norm1.weight": "encoders.0.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.layer_norm2.bias": "encoders.0.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.layer_norm2.weight": "encoders.0.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.mlp.fc1.bias": "encoders.0.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.mlp.fc1.weight": "encoders.0.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.mlp.fc2.bias": "encoders.0.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.mlp.fc2.weight": "encoders.0.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.k_proj.bias": "encoders.0.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.k_proj.weight": "encoders.0.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.out_proj.bias": "encoders.0.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.out_proj.weight": "encoders.0.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.q_proj.bias": "encoders.0.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.q_proj.weight": "encoders.0.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.v_proj.bias": "encoders.0.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.0.self_attn.v_proj.weight": "encoders.0.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.layer_norm1.bias": "encoders.1.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.layer_norm1.weight": "encoders.1.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.layer_norm2.bias": "encoders.1.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.layer_norm2.weight": "encoders.1.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.mlp.fc1.bias": "encoders.1.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.mlp.fc1.weight": "encoders.1.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.mlp.fc2.bias": "encoders.1.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.mlp.fc2.weight": "encoders.1.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.k_proj.bias": "encoders.1.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.k_proj.weight": "encoders.1.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.out_proj.bias": "encoders.1.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.out_proj.weight": "encoders.1.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.q_proj.bias": "encoders.1.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.q_proj.weight": "encoders.1.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.v_proj.bias": "encoders.1.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.1.self_attn.v_proj.weight": "encoders.1.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.layer_norm1.bias": "encoders.10.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.layer_norm1.weight": "encoders.10.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.layer_norm2.bias": "encoders.10.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.layer_norm2.weight": "encoders.10.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.mlp.fc1.bias": "encoders.10.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.mlp.fc1.weight": "encoders.10.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.mlp.fc2.bias": "encoders.10.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.mlp.fc2.weight": "encoders.10.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.k_proj.bias": "encoders.10.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.k_proj.weight": "encoders.10.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.out_proj.bias": "encoders.10.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.out_proj.weight": "encoders.10.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.q_proj.bias": "encoders.10.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.q_proj.weight": "encoders.10.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.v_proj.bias": "encoders.10.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.10.self_attn.v_proj.weight": "encoders.10.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.layer_norm1.bias": "encoders.11.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.layer_norm1.weight": "encoders.11.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.layer_norm2.bias": "encoders.11.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.layer_norm2.weight": "encoders.11.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.mlp.fc1.bias": "encoders.11.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.mlp.fc1.weight": "encoders.11.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.mlp.fc2.bias": "encoders.11.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.mlp.fc2.weight": "encoders.11.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.k_proj.bias": "encoders.11.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.k_proj.weight": "encoders.11.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.out_proj.bias": "encoders.11.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.out_proj.weight": "encoders.11.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.q_proj.bias": "encoders.11.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.q_proj.weight": "encoders.11.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.v_proj.bias": "encoders.11.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.11.self_attn.v_proj.weight": "encoders.11.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.layer_norm1.bias": "encoders.12.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.layer_norm1.weight": "encoders.12.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.layer_norm2.bias": "encoders.12.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.layer_norm2.weight": "encoders.12.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.mlp.fc1.bias": "encoders.12.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.mlp.fc1.weight": "encoders.12.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.mlp.fc2.bias": "encoders.12.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.mlp.fc2.weight": "encoders.12.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.k_proj.bias": "encoders.12.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.k_proj.weight": "encoders.12.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.out_proj.bias": "encoders.12.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.out_proj.weight": "encoders.12.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.q_proj.bias": "encoders.12.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.q_proj.weight": "encoders.12.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.v_proj.bias": "encoders.12.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.12.self_attn.v_proj.weight": "encoders.12.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.layer_norm1.bias": "encoders.13.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.layer_norm1.weight": "encoders.13.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.layer_norm2.bias": "encoders.13.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.layer_norm2.weight": "encoders.13.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.mlp.fc1.bias": "encoders.13.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.mlp.fc1.weight": "encoders.13.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.mlp.fc2.bias": "encoders.13.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.mlp.fc2.weight": "encoders.13.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.k_proj.bias": "encoders.13.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.k_proj.weight": "encoders.13.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.out_proj.bias": "encoders.13.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.out_proj.weight": "encoders.13.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.q_proj.bias": "encoders.13.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.q_proj.weight": "encoders.13.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.v_proj.bias": "encoders.13.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.13.self_attn.v_proj.weight": "encoders.13.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.layer_norm1.bias": "encoders.14.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.layer_norm1.weight": "encoders.14.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.layer_norm2.bias": "encoders.14.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.layer_norm2.weight": "encoders.14.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.mlp.fc1.bias": "encoders.14.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.mlp.fc1.weight": "encoders.14.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.mlp.fc2.bias": "encoders.14.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.mlp.fc2.weight": "encoders.14.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.k_proj.bias": "encoders.14.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.k_proj.weight": "encoders.14.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.out_proj.bias": "encoders.14.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.out_proj.weight": "encoders.14.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.q_proj.bias": "encoders.14.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.q_proj.weight": "encoders.14.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.v_proj.bias": "encoders.14.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.14.self_attn.v_proj.weight": "encoders.14.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.layer_norm1.bias": "encoders.15.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.layer_norm1.weight": "encoders.15.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.layer_norm2.bias": "encoders.15.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.layer_norm2.weight": "encoders.15.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.mlp.fc1.bias": "encoders.15.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.mlp.fc1.weight": "encoders.15.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.mlp.fc2.bias": "encoders.15.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.mlp.fc2.weight": "encoders.15.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.k_proj.bias": "encoders.15.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.k_proj.weight": "encoders.15.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.out_proj.bias": "encoders.15.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.out_proj.weight": "encoders.15.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.q_proj.bias": "encoders.15.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.q_proj.weight": "encoders.15.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.v_proj.bias": "encoders.15.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.15.self_attn.v_proj.weight": "encoders.15.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.layer_norm1.bias": "encoders.16.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.layer_norm1.weight": "encoders.16.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.layer_norm2.bias": "encoders.16.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.layer_norm2.weight": "encoders.16.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.mlp.fc1.bias": "encoders.16.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.mlp.fc1.weight": "encoders.16.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.mlp.fc2.bias": "encoders.16.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.mlp.fc2.weight": "encoders.16.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.k_proj.bias": "encoders.16.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.k_proj.weight": "encoders.16.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.out_proj.bias": "encoders.16.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.out_proj.weight": "encoders.16.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.q_proj.bias": "encoders.16.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.q_proj.weight": "encoders.16.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.v_proj.bias": "encoders.16.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.16.self_attn.v_proj.weight": "encoders.16.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.layer_norm1.bias": "encoders.17.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.layer_norm1.weight": "encoders.17.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.layer_norm2.bias": "encoders.17.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.layer_norm2.weight": "encoders.17.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.mlp.fc1.bias": "encoders.17.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.mlp.fc1.weight": "encoders.17.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.mlp.fc2.bias": "encoders.17.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.mlp.fc2.weight": "encoders.17.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.k_proj.bias": "encoders.17.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.k_proj.weight": "encoders.17.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.out_proj.bias": "encoders.17.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.out_proj.weight": "encoders.17.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.q_proj.bias": "encoders.17.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.q_proj.weight": "encoders.17.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.v_proj.bias": "encoders.17.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.17.self_attn.v_proj.weight": "encoders.17.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.layer_norm1.bias": "encoders.18.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.layer_norm1.weight": "encoders.18.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.layer_norm2.bias": "encoders.18.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.layer_norm2.weight": "encoders.18.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.mlp.fc1.bias": "encoders.18.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.mlp.fc1.weight": "encoders.18.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.mlp.fc2.bias": "encoders.18.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.mlp.fc2.weight": "encoders.18.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.k_proj.bias": "encoders.18.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.k_proj.weight": "encoders.18.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.out_proj.bias": "encoders.18.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.out_proj.weight": "encoders.18.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.q_proj.bias": "encoders.18.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.q_proj.weight": "encoders.18.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.v_proj.bias": "encoders.18.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.18.self_attn.v_proj.weight": "encoders.18.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.layer_norm1.bias": "encoders.19.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.layer_norm1.weight": "encoders.19.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.layer_norm2.bias": "encoders.19.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.layer_norm2.weight": "encoders.19.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.mlp.fc1.bias": "encoders.19.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.mlp.fc1.weight": "encoders.19.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.mlp.fc2.bias": "encoders.19.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.mlp.fc2.weight": "encoders.19.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.k_proj.bias": "encoders.19.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.k_proj.weight": "encoders.19.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.out_proj.bias": "encoders.19.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.out_proj.weight": "encoders.19.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.q_proj.bias": "encoders.19.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.q_proj.weight": "encoders.19.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.v_proj.bias": "encoders.19.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.19.self_attn.v_proj.weight": "encoders.19.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.layer_norm1.bias": "encoders.2.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.layer_norm1.weight": "encoders.2.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.layer_norm2.bias": "encoders.2.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.layer_norm2.weight": "encoders.2.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.mlp.fc1.bias": "encoders.2.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.mlp.fc1.weight": "encoders.2.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.mlp.fc2.bias": "encoders.2.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.mlp.fc2.weight": "encoders.2.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.k_proj.bias": "encoders.2.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.k_proj.weight": "encoders.2.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.out_proj.bias": "encoders.2.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.out_proj.weight": "encoders.2.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.q_proj.bias": "encoders.2.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.q_proj.weight": "encoders.2.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.v_proj.bias": "encoders.2.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.2.self_attn.v_proj.weight": "encoders.2.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.layer_norm1.bias": "encoders.20.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.layer_norm1.weight": "encoders.20.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.layer_norm2.bias": "encoders.20.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.layer_norm2.weight": "encoders.20.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.mlp.fc1.bias": "encoders.20.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.mlp.fc1.weight": "encoders.20.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.mlp.fc2.bias": "encoders.20.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.mlp.fc2.weight": "encoders.20.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.k_proj.bias": "encoders.20.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.k_proj.weight": "encoders.20.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.out_proj.bias": "encoders.20.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.out_proj.weight": "encoders.20.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.q_proj.bias": "encoders.20.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.q_proj.weight": "encoders.20.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.v_proj.bias": "encoders.20.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.20.self_attn.v_proj.weight": "encoders.20.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.layer_norm1.bias": "encoders.21.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.layer_norm1.weight": "encoders.21.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.layer_norm2.bias": "encoders.21.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.layer_norm2.weight": "encoders.21.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.mlp.fc1.bias": "encoders.21.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.mlp.fc1.weight": "encoders.21.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.mlp.fc2.bias": "encoders.21.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.mlp.fc2.weight": "encoders.21.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.k_proj.bias": "encoders.21.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.k_proj.weight": "encoders.21.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.out_proj.bias": "encoders.21.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.out_proj.weight": "encoders.21.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.q_proj.bias": "encoders.21.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.q_proj.weight": "encoders.21.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.v_proj.bias": "encoders.21.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.21.self_attn.v_proj.weight": "encoders.21.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.layer_norm1.bias": "encoders.22.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.layer_norm1.weight": "encoders.22.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.layer_norm2.bias": "encoders.22.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.layer_norm2.weight": "encoders.22.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.mlp.fc1.bias": "encoders.22.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.mlp.fc1.weight": "encoders.22.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.mlp.fc2.bias": "encoders.22.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.mlp.fc2.weight": "encoders.22.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.k_proj.bias": "encoders.22.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.k_proj.weight": "encoders.22.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.out_proj.bias": "encoders.22.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.out_proj.weight": "encoders.22.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.q_proj.bias": "encoders.22.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.q_proj.weight": "encoders.22.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.v_proj.bias": "encoders.22.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.22.self_attn.v_proj.weight": "encoders.22.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.layer_norm1.bias": "encoders.23.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.layer_norm1.weight": "encoders.23.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.layer_norm2.bias": "encoders.23.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.layer_norm2.weight": "encoders.23.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.mlp.fc1.bias": "encoders.23.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.mlp.fc1.weight": "encoders.23.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.mlp.fc2.bias": "encoders.23.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.mlp.fc2.weight": "encoders.23.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.k_proj.bias": "encoders.23.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.k_proj.weight": "encoders.23.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.out_proj.bias": "encoders.23.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.out_proj.weight": "encoders.23.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.q_proj.bias": "encoders.23.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.q_proj.weight": "encoders.23.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.v_proj.bias": "encoders.23.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.23.self_attn.v_proj.weight": "encoders.23.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.layer_norm1.bias": "encoders.24.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.layer_norm1.weight": "encoders.24.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.layer_norm2.bias": "encoders.24.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.layer_norm2.weight": "encoders.24.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.mlp.fc1.bias": "encoders.24.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.mlp.fc1.weight": "encoders.24.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.mlp.fc2.bias": "encoders.24.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.mlp.fc2.weight": "encoders.24.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.k_proj.bias": "encoders.24.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.k_proj.weight": "encoders.24.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.out_proj.bias": "encoders.24.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.out_proj.weight": "encoders.24.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.q_proj.bias": "encoders.24.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.q_proj.weight": "encoders.24.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.v_proj.bias": "encoders.24.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.24.self_attn.v_proj.weight": "encoders.24.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.layer_norm1.bias": "encoders.25.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.layer_norm1.weight": "encoders.25.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.layer_norm2.bias": "encoders.25.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.layer_norm2.weight": "encoders.25.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.mlp.fc1.bias": "encoders.25.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.mlp.fc1.weight": "encoders.25.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.mlp.fc2.bias": "encoders.25.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.mlp.fc2.weight": "encoders.25.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.k_proj.bias": "encoders.25.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.k_proj.weight": "encoders.25.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.out_proj.bias": "encoders.25.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.out_proj.weight": "encoders.25.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.q_proj.bias": "encoders.25.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.q_proj.weight": "encoders.25.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.v_proj.bias": "encoders.25.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.25.self_attn.v_proj.weight": "encoders.25.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.layer_norm1.bias": "encoders.26.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.layer_norm1.weight": "encoders.26.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.layer_norm2.bias": "encoders.26.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.layer_norm2.weight": "encoders.26.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.mlp.fc1.bias": "encoders.26.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.mlp.fc1.weight": "encoders.26.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.mlp.fc2.bias": "encoders.26.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.mlp.fc2.weight": "encoders.26.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.k_proj.bias": "encoders.26.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.k_proj.weight": "encoders.26.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.out_proj.bias": "encoders.26.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.out_proj.weight": "encoders.26.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.q_proj.bias": "encoders.26.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.q_proj.weight": "encoders.26.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.v_proj.bias": "encoders.26.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.26.self_attn.v_proj.weight": "encoders.26.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.layer_norm1.bias": "encoders.27.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.layer_norm1.weight": "encoders.27.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.layer_norm2.bias": "encoders.27.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.layer_norm2.weight": "encoders.27.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.mlp.fc1.bias": "encoders.27.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.mlp.fc1.weight": "encoders.27.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.mlp.fc2.bias": "encoders.27.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.mlp.fc2.weight": "encoders.27.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.k_proj.bias": "encoders.27.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.k_proj.weight": "encoders.27.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.out_proj.bias": "encoders.27.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.out_proj.weight": "encoders.27.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.q_proj.bias": "encoders.27.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.q_proj.weight": "encoders.27.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.v_proj.bias": "encoders.27.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.27.self_attn.v_proj.weight": "encoders.27.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.layer_norm1.bias": "encoders.28.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.layer_norm1.weight": "encoders.28.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.layer_norm2.bias": "encoders.28.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.layer_norm2.weight": "encoders.28.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.mlp.fc1.bias": "encoders.28.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.mlp.fc1.weight": "encoders.28.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.mlp.fc2.bias": "encoders.28.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.mlp.fc2.weight": "encoders.28.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.k_proj.bias": "encoders.28.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.k_proj.weight": "encoders.28.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.out_proj.bias": "encoders.28.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.out_proj.weight": "encoders.28.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.q_proj.bias": "encoders.28.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.q_proj.weight": "encoders.28.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.v_proj.bias": "encoders.28.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.28.self_attn.v_proj.weight": "encoders.28.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.layer_norm1.bias": "encoders.29.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.layer_norm1.weight": "encoders.29.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.layer_norm2.bias": "encoders.29.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.layer_norm2.weight": "encoders.29.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.mlp.fc1.bias": "encoders.29.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.mlp.fc1.weight": "encoders.29.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.mlp.fc2.bias": "encoders.29.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.mlp.fc2.weight": "encoders.29.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.k_proj.bias": "encoders.29.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.k_proj.weight": "encoders.29.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.out_proj.bias": "encoders.29.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.out_proj.weight": "encoders.29.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.q_proj.bias": "encoders.29.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.q_proj.weight": "encoders.29.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.v_proj.bias": "encoders.29.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.29.self_attn.v_proj.weight": "encoders.29.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.layer_norm1.bias": "encoders.3.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.layer_norm1.weight": "encoders.3.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.layer_norm2.bias": "encoders.3.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.layer_norm2.weight": "encoders.3.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.mlp.fc1.bias": "encoders.3.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.mlp.fc1.weight": "encoders.3.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.mlp.fc2.bias": "encoders.3.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.mlp.fc2.weight": "encoders.3.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.k_proj.bias": "encoders.3.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.k_proj.weight": "encoders.3.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.out_proj.bias": "encoders.3.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.out_proj.weight": "encoders.3.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.q_proj.bias": "encoders.3.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.q_proj.weight": "encoders.3.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.v_proj.bias": "encoders.3.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.3.self_attn.v_proj.weight": "encoders.3.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.layer_norm1.bias": "encoders.30.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.layer_norm1.weight": "encoders.30.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.layer_norm2.bias": "encoders.30.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.layer_norm2.weight": "encoders.30.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.mlp.fc1.bias": "encoders.30.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.mlp.fc1.weight": "encoders.30.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.mlp.fc2.bias": "encoders.30.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.mlp.fc2.weight": "encoders.30.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.k_proj.bias": "encoders.30.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.k_proj.weight": "encoders.30.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.out_proj.bias": "encoders.30.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.out_proj.weight": "encoders.30.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.q_proj.bias": "encoders.30.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.q_proj.weight": "encoders.30.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.v_proj.bias": "encoders.30.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.30.self_attn.v_proj.weight": "encoders.30.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.layer_norm1.bias": "encoders.31.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.layer_norm1.weight": "encoders.31.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.layer_norm2.bias": "encoders.31.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.layer_norm2.weight": "encoders.31.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.mlp.fc1.bias": "encoders.31.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.mlp.fc1.weight": "encoders.31.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.mlp.fc2.bias": "encoders.31.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.mlp.fc2.weight": "encoders.31.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.k_proj.bias": "encoders.31.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.k_proj.weight": "encoders.31.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.out_proj.bias": "encoders.31.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.out_proj.weight": "encoders.31.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.q_proj.bias": "encoders.31.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.q_proj.weight": "encoders.31.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.v_proj.bias": "encoders.31.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.31.self_attn.v_proj.weight": "encoders.31.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.layer_norm1.bias": "encoders.4.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.layer_norm1.weight": "encoders.4.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.layer_norm2.bias": "encoders.4.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.layer_norm2.weight": "encoders.4.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.mlp.fc1.bias": "encoders.4.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.mlp.fc1.weight": "encoders.4.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.mlp.fc2.bias": "encoders.4.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.mlp.fc2.weight": "encoders.4.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.k_proj.bias": "encoders.4.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.k_proj.weight": "encoders.4.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.out_proj.bias": "encoders.4.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.out_proj.weight": "encoders.4.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.q_proj.bias": "encoders.4.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.q_proj.weight": "encoders.4.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.v_proj.bias": "encoders.4.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.4.self_attn.v_proj.weight": "encoders.4.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.layer_norm1.bias": "encoders.5.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.layer_norm1.weight": "encoders.5.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.layer_norm2.bias": "encoders.5.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.layer_norm2.weight": "encoders.5.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.mlp.fc1.bias": "encoders.5.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.mlp.fc1.weight": "encoders.5.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.mlp.fc2.bias": "encoders.5.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.mlp.fc2.weight": "encoders.5.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.k_proj.bias": "encoders.5.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.k_proj.weight": "encoders.5.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.out_proj.bias": "encoders.5.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.out_proj.weight": "encoders.5.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.q_proj.bias": "encoders.5.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.q_proj.weight": "encoders.5.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.v_proj.bias": "encoders.5.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.5.self_attn.v_proj.weight": "encoders.5.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.layer_norm1.bias": "encoders.6.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.layer_norm1.weight": "encoders.6.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.layer_norm2.bias": "encoders.6.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.layer_norm2.weight": "encoders.6.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.mlp.fc1.bias": "encoders.6.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.mlp.fc1.weight": "encoders.6.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.mlp.fc2.bias": "encoders.6.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.mlp.fc2.weight": "encoders.6.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.k_proj.bias": "encoders.6.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.k_proj.weight": "encoders.6.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.out_proj.bias": "encoders.6.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.out_proj.weight": "encoders.6.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.q_proj.bias": "encoders.6.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.q_proj.weight": "encoders.6.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.v_proj.bias": "encoders.6.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.6.self_attn.v_proj.weight": "encoders.6.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.layer_norm1.bias": "encoders.7.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.layer_norm1.weight": "encoders.7.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.layer_norm2.bias": "encoders.7.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.layer_norm2.weight": "encoders.7.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.mlp.fc1.bias": "encoders.7.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.mlp.fc1.weight": "encoders.7.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.mlp.fc2.bias": "encoders.7.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.mlp.fc2.weight": "encoders.7.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.k_proj.bias": "encoders.7.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.k_proj.weight": "encoders.7.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.out_proj.bias": "encoders.7.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.out_proj.weight": "encoders.7.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.q_proj.bias": "encoders.7.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.q_proj.weight": "encoders.7.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.v_proj.bias": "encoders.7.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.7.self_attn.v_proj.weight": "encoders.7.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.layer_norm1.bias": "encoders.8.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.layer_norm1.weight": "encoders.8.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.layer_norm2.bias": "encoders.8.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.layer_norm2.weight": "encoders.8.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.mlp.fc1.bias": "encoders.8.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.mlp.fc1.weight": "encoders.8.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.mlp.fc2.bias": "encoders.8.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.mlp.fc2.weight": "encoders.8.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.k_proj.bias": "encoders.8.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.k_proj.weight": "encoders.8.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.out_proj.bias": "encoders.8.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.out_proj.weight": "encoders.8.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.q_proj.bias": "encoders.8.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.q_proj.weight": "encoders.8.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.v_proj.bias": "encoders.8.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.8.self_attn.v_proj.weight": "encoders.8.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.layer_norm1.bias": "encoders.9.layer_norm1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.layer_norm1.weight": "encoders.9.layer_norm1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.layer_norm2.bias": "encoders.9.layer_norm2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.layer_norm2.weight": "encoders.9.layer_norm2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.mlp.fc1.bias": "encoders.9.fc1.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.mlp.fc1.weight": "encoders.9.fc1.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.mlp.fc2.bias": "encoders.9.fc2.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.mlp.fc2.weight": "encoders.9.fc2.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.k_proj.bias": "encoders.9.attn.to_k.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.k_proj.weight": "encoders.9.attn.to_k.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.out_proj.bias": "encoders.9.attn.to_out.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.out_proj.weight": "encoders.9.attn.to_out.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.q_proj.bias": "encoders.9.attn.to_q.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.q_proj.weight": "encoders.9.attn.to_q.weight", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.v_proj.bias": "encoders.9.attn.to_v.bias", "text_encoders.clip_g.transformer.text_model.encoder.layers.9.self_attn.v_proj.weight": "encoders.9.attn.to_v.weight", "text_encoders.clip_g.transformer.text_model.final_layer_norm.bias": "final_layer_norm.bias", "text_encoders.clip_g.transformer.text_model.final_layer_norm.weight": "final_layer_norm.weight", "text_encoders.clip_g.transformer.text_projection.weight": "text_projection.weight", } state_dict_ = {} for name in state_dict: if name in rename_dict: param = state_dict[name] if name == "text_encoders.clip_g.transformer.text_model.embeddings.position_embedding.weight": param = param.reshape((1, param.shape[0], param.shape[1])) state_dict_[rename_dict[name]] = param return state_dict_ class SD3TextEncoder3StateDictConverter(): def __init__(self): pass def from_diffusers(self, state_dict): state_dict_ = state_dict return state_dict_ def from_civitai(self, state_dict): prefix = "text_encoders.t5xxl.transformer." state_dict_ = {name[len(prefix):]: param for name, param in state_dict.items() if name.startswith(prefix)} if len(state_dict_) > 0: return self.from_diffusers(state_dict_) name_list = [ "encoder.block.0.layer.0.SelfAttention.k.weight", "encoder.block.0.layer.0.SelfAttention.o.weight", "encoder.block.0.layer.0.SelfAttention.q.weight", "encoder.block.0.layer.0.SelfAttention.relative_attention_bias.weight", "encoder.block.0.layer.0.SelfAttention.v.weight", "encoder.block.0.layer.0.layer_norm.weight", "encoder.block.0.layer.1.DenseReluDense.wi_0.weight", "encoder.block.0.layer.1.DenseReluDense.wi_1.weight", "encoder.block.0.layer.1.DenseReluDense.wo.weight", "encoder.block.0.layer.1.layer_norm.weight", "encoder.block.1.layer.0.SelfAttention.k.weight", "encoder.block.1.layer.0.SelfAttention.o.weight", "encoder.block.1.layer.0.SelfAttention.q.weight", "encoder.block.1.layer.0.SelfAttention.v.weight", "encoder.block.1.layer.0.layer_norm.weight", "encoder.block.1.layer.1.DenseReluDense.wi_0.weight", "encoder.block.1.layer.1.DenseReluDense.wi_1.weight", "encoder.block.1.layer.1.DenseReluDense.wo.weight", "encoder.block.1.layer.1.layer_norm.weight", "encoder.block.10.layer.0.SelfAttention.k.weight", "encoder.block.10.layer.0.SelfAttention.o.weight", "encoder.block.10.layer.0.SelfAttention.q.weight", "encoder.block.10.layer.0.SelfAttention.v.weight", "encoder.block.10.layer.0.layer_norm.weight", "encoder.block.10.layer.1.DenseReluDense.wi_0.weight", "encoder.block.10.layer.1.DenseReluDense.wi_1.weight", "encoder.block.10.layer.1.DenseReluDense.wo.weight", "encoder.block.10.layer.1.layer_norm.weight", "encoder.block.11.layer.0.SelfAttention.k.weight", "encoder.block.11.layer.0.SelfAttention.o.weight", "encoder.block.11.layer.0.SelfAttention.q.weight", "encoder.block.11.layer.0.SelfAttention.v.weight", "encoder.block.11.layer.0.layer_norm.weight", "encoder.block.11.layer.1.DenseReluDense.wi_0.weight", "encoder.block.11.layer.1.DenseReluDense.wi_1.weight", "encoder.block.11.layer.1.DenseReluDense.wo.weight", "encoder.block.11.layer.1.layer_norm.weight", "encoder.block.12.layer.0.SelfAttention.k.weight", "encoder.block.12.layer.0.SelfAttention.o.weight", "encoder.block.12.layer.0.SelfAttention.q.weight", "encoder.block.12.layer.0.SelfAttention.v.weight", "encoder.block.12.layer.0.layer_norm.weight", "encoder.block.12.layer.1.DenseReluDense.wi_0.weight", "encoder.block.12.layer.1.DenseReluDense.wi_1.weight", "encoder.block.12.layer.1.DenseReluDense.wo.weight", "encoder.block.12.layer.1.layer_norm.weight", "encoder.block.13.layer.0.SelfAttention.k.weight", "encoder.block.13.layer.0.SelfAttention.o.weight", "encoder.block.13.layer.0.SelfAttention.q.weight", "encoder.block.13.layer.0.SelfAttention.v.weight", "encoder.block.13.layer.0.layer_norm.weight", "encoder.block.13.layer.1.DenseReluDense.wi_0.weight", "encoder.block.13.layer.1.DenseReluDense.wi_1.weight", "encoder.block.13.layer.1.DenseReluDense.wo.weight", "encoder.block.13.layer.1.layer_norm.weight", "encoder.block.14.layer.0.SelfAttention.k.weight", "encoder.block.14.layer.0.SelfAttention.o.weight", "encoder.block.14.layer.0.SelfAttention.q.weight", "encoder.block.14.layer.0.SelfAttention.v.weight", "encoder.block.14.layer.0.layer_norm.weight", "encoder.block.14.layer.1.DenseReluDense.wi_0.weight", "encoder.block.14.layer.1.DenseReluDense.wi_1.weight", "encoder.block.14.layer.1.DenseReluDense.wo.weight", "encoder.block.14.layer.1.layer_norm.weight", "encoder.block.15.layer.0.SelfAttention.k.weight", "encoder.block.15.layer.0.SelfAttention.o.weight", "encoder.block.15.layer.0.SelfAttention.q.weight", "encoder.block.15.layer.0.SelfAttention.v.weight", "encoder.block.15.layer.0.layer_norm.weight", "encoder.block.15.layer.1.DenseReluDense.wi_0.weight", "encoder.block.15.layer.1.DenseReluDense.wi_1.weight", "encoder.block.15.layer.1.DenseReluDense.wo.weight", "encoder.block.15.layer.1.layer_norm.weight", "encoder.block.16.layer.0.SelfAttention.k.weight", "encoder.block.16.layer.0.SelfAttention.o.weight", "encoder.block.16.layer.0.SelfAttention.q.weight", "encoder.block.16.layer.0.SelfAttention.v.weight", "encoder.block.16.layer.0.layer_norm.weight", "encoder.block.16.layer.1.DenseReluDense.wi_0.weight", "encoder.block.16.layer.1.DenseReluDense.wi_1.weight", "encoder.block.16.layer.1.DenseReluDense.wo.weight", "encoder.block.16.layer.1.layer_norm.weight", "encoder.block.17.layer.0.SelfAttention.k.weight", "encoder.block.17.layer.0.SelfAttention.o.weight", "encoder.block.17.layer.0.SelfAttention.q.weight", "encoder.block.17.layer.0.SelfAttention.v.weight", "encoder.block.17.layer.0.layer_norm.weight", "encoder.block.17.layer.1.DenseReluDense.wi_0.weight", "encoder.block.17.layer.1.DenseReluDense.wi_1.weight", "encoder.block.17.layer.1.DenseReluDense.wo.weight", "encoder.block.17.layer.1.layer_norm.weight", "encoder.block.18.layer.0.SelfAttention.k.weight", "encoder.block.18.layer.0.SelfAttention.o.weight", "encoder.block.18.layer.0.SelfAttention.q.weight", "encoder.block.18.layer.0.SelfAttention.v.weight", "encoder.block.18.layer.0.layer_norm.weight", "encoder.block.18.layer.1.DenseReluDense.wi_0.weight", "encoder.block.18.layer.1.DenseReluDense.wi_1.weight", "encoder.block.18.layer.1.DenseReluDense.wo.weight", "encoder.block.18.layer.1.layer_norm.weight", "encoder.block.19.layer.0.SelfAttention.k.weight", "encoder.block.19.layer.0.SelfAttention.o.weight", "encoder.block.19.layer.0.SelfAttention.q.weight", "encoder.block.19.layer.0.SelfAttention.v.weight", "encoder.block.19.layer.0.layer_norm.weight", "encoder.block.19.layer.1.DenseReluDense.wi_0.weight", "encoder.block.19.layer.1.DenseReluDense.wi_1.weight", "encoder.block.19.layer.1.DenseReluDense.wo.weight", "encoder.block.19.layer.1.layer_norm.weight", "encoder.block.2.layer.0.SelfAttention.k.weight", "encoder.block.2.layer.0.SelfAttention.o.weight", "encoder.block.2.layer.0.SelfAttention.q.weight", "encoder.block.2.layer.0.SelfAttention.v.weight", "encoder.block.2.layer.0.layer_norm.weight", "encoder.block.2.layer.1.DenseReluDense.wi_0.weight", "encoder.block.2.layer.1.DenseReluDense.wi_1.weight", "encoder.block.2.layer.1.DenseReluDense.wo.weight", "encoder.block.2.layer.1.layer_norm.weight", "encoder.block.20.layer.0.SelfAttention.k.weight", "encoder.block.20.layer.0.SelfAttention.o.weight", "encoder.block.20.layer.0.SelfAttention.q.weight", "encoder.block.20.layer.0.SelfAttention.v.weight", "encoder.block.20.layer.0.layer_norm.weight", "encoder.block.20.layer.1.DenseReluDense.wi_0.weight", "encoder.block.20.layer.1.DenseReluDense.wi_1.weight", "encoder.block.20.layer.1.DenseReluDense.wo.weight", "encoder.block.20.layer.1.layer_norm.weight", "encoder.block.21.layer.0.SelfAttention.k.weight", "encoder.block.21.layer.0.SelfAttention.o.weight", "encoder.block.21.layer.0.SelfAttention.q.weight", "encoder.block.21.layer.0.SelfAttention.v.weight", "encoder.block.21.layer.0.layer_norm.weight", "encoder.block.21.layer.1.DenseReluDense.wi_0.weight", "encoder.block.21.layer.1.DenseReluDense.wi_1.weight", "encoder.block.21.layer.1.DenseReluDense.wo.weight", "encoder.block.21.layer.1.layer_norm.weight", "encoder.block.22.layer.0.SelfAttention.k.weight", "encoder.block.22.layer.0.SelfAttention.o.weight", "encoder.block.22.layer.0.SelfAttention.q.weight", "encoder.block.22.layer.0.SelfAttention.v.weight", "encoder.block.22.layer.0.layer_norm.weight", "encoder.block.22.layer.1.DenseReluDense.wi_0.weight", "encoder.block.22.layer.1.DenseReluDense.wi_1.weight", "encoder.block.22.layer.1.DenseReluDense.wo.weight", "encoder.block.22.layer.1.layer_norm.weight", "encoder.block.23.layer.0.SelfAttention.k.weight", "encoder.block.23.layer.0.SelfAttention.o.weight", "encoder.block.23.layer.0.SelfAttention.q.weight", "encoder.block.23.layer.0.SelfAttention.v.weight", "encoder.block.23.layer.0.layer_norm.weight", "encoder.block.23.layer.1.DenseReluDense.wi_0.weight", "encoder.block.23.layer.1.DenseReluDense.wi_1.weight", "encoder.block.23.layer.1.DenseReluDense.wo.weight", "encoder.block.23.layer.1.layer_norm.weight", "encoder.block.3.layer.0.SelfAttention.k.weight", "encoder.block.3.layer.0.SelfAttention.o.weight", "encoder.block.3.layer.0.SelfAttention.q.weight", "encoder.block.3.layer.0.SelfAttention.v.weight", "encoder.block.3.layer.0.layer_norm.weight", "encoder.block.3.layer.1.DenseReluDense.wi_0.weight", "encoder.block.3.layer.1.DenseReluDense.wi_1.weight", "encoder.block.3.layer.1.DenseReluDense.wo.weight", "encoder.block.3.layer.1.layer_norm.weight", "encoder.block.4.layer.0.SelfAttention.k.weight", "encoder.block.4.layer.0.SelfAttention.o.weight", "encoder.block.4.layer.0.SelfAttention.q.weight", "encoder.block.4.layer.0.SelfAttention.v.weight", "encoder.block.4.layer.0.layer_norm.weight", "encoder.block.4.layer.1.DenseReluDense.wi_0.weight", "encoder.block.4.layer.1.DenseReluDense.wi_1.weight", "encoder.block.4.layer.1.DenseReluDense.wo.weight", "encoder.block.4.layer.1.layer_norm.weight", "encoder.block.5.layer.0.SelfAttention.k.weight", "encoder.block.5.layer.0.SelfAttention.o.weight", "encoder.block.5.layer.0.SelfAttention.q.weight", "encoder.block.5.layer.0.SelfAttention.v.weight", "encoder.block.5.layer.0.layer_norm.weight", "encoder.block.5.layer.1.DenseReluDense.wi_0.weight", "encoder.block.5.layer.1.DenseReluDense.wi_1.weight", "encoder.block.5.layer.1.DenseReluDense.wo.weight", "encoder.block.5.layer.1.layer_norm.weight", "encoder.block.6.layer.0.SelfAttention.k.weight", "encoder.block.6.layer.0.SelfAttention.o.weight", "encoder.block.6.layer.0.SelfAttention.q.weight", "encoder.block.6.layer.0.SelfAttention.v.weight", "encoder.block.6.layer.0.layer_norm.weight", "encoder.block.6.layer.1.DenseReluDense.wi_0.weight", "encoder.block.6.layer.1.DenseReluDense.wi_1.weight", "encoder.block.6.layer.1.DenseReluDense.wo.weight", "encoder.block.6.layer.1.layer_norm.weight", "encoder.block.7.layer.0.SelfAttention.k.weight", "encoder.block.7.layer.0.SelfAttention.o.weight", "encoder.block.7.layer.0.SelfAttention.q.weight", "encoder.block.7.layer.0.SelfAttention.v.weight", "encoder.block.7.layer.0.layer_norm.weight", "encoder.block.7.layer.1.DenseReluDense.wi_0.weight", "encoder.block.7.layer.1.DenseReluDense.wi_1.weight", "encoder.block.7.layer.1.DenseReluDense.wo.weight", "encoder.block.7.layer.1.layer_norm.weight", "encoder.block.8.layer.0.SelfAttention.k.weight", "encoder.block.8.layer.0.SelfAttention.o.weight", "encoder.block.8.layer.0.SelfAttention.q.weight", "encoder.block.8.layer.0.SelfAttention.v.weight", "encoder.block.8.layer.0.layer_norm.weight", "encoder.block.8.layer.1.DenseReluDense.wi_0.weight", "encoder.block.8.layer.1.DenseReluDense.wi_1.weight", "encoder.block.8.layer.1.DenseReluDense.wo.weight", "encoder.block.8.layer.1.layer_norm.weight", "encoder.block.9.layer.0.SelfAttention.k.weight", "encoder.block.9.layer.0.SelfAttention.o.weight", "encoder.block.9.layer.0.SelfAttention.q.weight", "encoder.block.9.layer.0.SelfAttention.v.weight", "encoder.block.9.layer.0.layer_norm.weight", "encoder.block.9.layer.1.DenseReluDense.wi_0.weight", "encoder.block.9.layer.1.DenseReluDense.wi_1.weight", "encoder.block.9.layer.1.DenseReluDense.wo.weight", "encoder.block.9.layer.1.layer_norm.weight", "encoder.embed_tokens.weight", "encoder.final_layer_norm.weight", "shared.weight", ] state_dict_ = {} for name, param in state_dict.items(): if name in name_list: state_dict_[name] = param return state_dict_