{ "architectures": [ "CLIPTextModel" ], "attention_dropout": 0.0, "bos_token_id": 0, "dropout": 0.0, "dtype": "bfloat16", "eos_token_id": 2, "hidden_act": "quick_gelu", "hidden_size": 768, "initializer_factor": 1.0, "initializer_range": 0.02, "intermediate_size": 3072, "layer_norm_eps": 1e-05, "max_position_embeddings": 77, "model_type": "clip_text_model", "num_attention_heads": 12, "num_hidden_layers": 12, "pad_token_id": 1, "projection_dim": 512, "quantization_config": { "_load_in_4bit": true, "_load_in_8bit": false, "bnb_4bit_compute_dtype": "bfloat16", "bnb_4bit_quant_storage": "uint8", "bnb_4bit_quant_type": "nf4", "bnb_4bit_use_double_quant": true, "llm_int8_enable_fp32_cpu_offload": false, "llm_int8_has_fp16_weight": false, "llm_int8_skip_modules": [ "down_blocks.0.attentions.0.proj_in", "down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_q", "down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_k", "down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_v", "down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_out.0", "down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_q", "down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_k", "down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_v", "down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_out.0", "down_blocks.0.attentions.0.transformer_blocks.0.ff.net.0.proj", "down_blocks.0.attentions.0.transformer_blocks.0.ff.net.2", "down_blocks.0.attentions.0.proj_out", "down_blocks.0.attentions.1.proj_in", "down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_q", "down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_k", "down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_v", "down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_out.0", "down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_q", "down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_k", "down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_v", "down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_out.0", "down_blocks.0.attentions.1.transformer_blocks.0.ff.net.0.proj", "down_blocks.0.attentions.1.transformer_blocks.0.ff.net.2", "down_blocks.0.attentions.1.proj_out", "down_blocks.0.resnets.0.conv1", "down_blocks.0.resnets.0.time_emb_proj", "down_blocks.0.resnets.0.conv2", "down_blocks.0.resnets.1.conv1", "down_blocks.0.resnets.1.time_emb_proj", "down_blocks.0.resnets.1.conv2", "down_blocks.0.downsamplers.0.conv", "up_blocks.3.attentions.0.proj_in", "up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_q", "up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_k", "up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_v", "up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_out.0", "up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_q", "up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_k", "up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_v", "up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_out.0", "up_blocks.3.attentions.0.transformer_blocks.0.ff.net.0.proj", "up_blocks.3.attentions.0.transformer_blocks.0.ff.net.2", "up_blocks.3.attentions.0.proj_out", "up_blocks.3.attentions.1.proj_in", "up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_q", "up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_k", "up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_v", "up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_out.0", "up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_q", "up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_k", "up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_v", "up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_out.0", "up_blocks.3.attentions.1.transformer_blocks.0.ff.net.0.proj", "up_blocks.3.attentions.1.transformer_blocks.0.ff.net.2", "up_blocks.3.attentions.1.proj_out", "up_blocks.3.attentions.2.proj_in", "up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_q", "up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_k", "up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_v", "up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_out.0", "up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_q", "up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_k", "up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_v", "up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_out.0", "up_blocks.3.attentions.2.transformer_blocks.0.ff.net.0.proj", "up_blocks.3.attentions.2.transformer_blocks.0.ff.net.2", "up_blocks.3.attentions.2.proj_out", "up_blocks.3.resnets.0.conv1", "up_blocks.3.resnets.0.time_emb_proj", "up_blocks.3.resnets.0.conv2", "up_blocks.3.resnets.0.conv_shortcut", "up_blocks.3.resnets.1.conv1", "up_blocks.3.resnets.1.time_emb_proj", "up_blocks.3.resnets.1.conv2", "up_blocks.3.resnets.1.conv_shortcut", "up_blocks.3.resnets.2.conv1", "up_blocks.3.resnets.2.time_emb_proj", "up_blocks.3.resnets.2.conv2", "up_blocks.3.resnets.2.conv_shortcut", "mid_block.attentions.0.proj_in", "mid_block.attentions.0.transformer_blocks.0.attn1.to_q", "mid_block.attentions.0.transformer_blocks.0.attn1.to_k", "mid_block.attentions.0.transformer_blocks.0.attn1.to_v", "mid_block.attentions.0.transformer_blocks.0.attn1.to_out.0", "mid_block.attentions.0.transformer_blocks.0.attn2.to_q", "mid_block.attentions.0.transformer_blocks.0.attn2.to_k", "mid_block.attentions.0.transformer_blocks.0.attn2.to_v", "mid_block.attentions.0.transformer_blocks.0.attn2.to_out.0", "mid_block.attentions.0.transformer_blocks.0.ff.net.0.proj", "mid_block.attentions.0.transformer_blocks.0.ff.net.2", "mid_block.attentions.0.proj_out", "mid_block.resnets.0.conv1", "mid_block.resnets.0.time_emb_proj", "mid_block.resnets.0.conv2", "mid_block.resnets.1.conv1", "mid_block.resnets.1.time_emb_proj", "mid_block.resnets.1.conv2", "conv_in", "conv_out" ], "llm_int8_threshold": 6.0, "load_in_4bit": true, "load_in_8bit": false, "quant_method": "bitsandbytes" }, "transformers_version": "4.57.6", "vocab_size": 49408 }