| { |
| "_class_name": "AutoencoderKL", |
| "_diffusers_version": "0.36.0.dev0", |
| "_name_or_path": "./tmp_model/vae", |
| "act_fn": "silu", |
| "block_out_channels": [ |
| 128, |
| 256, |
| 512, |
| 512 |
| ], |
| "down_block_types": [ |
| "DownEncoderBlock2D", |
| "DownEncoderBlock2D", |
| "DownEncoderBlock2D", |
| "DownEncoderBlock2D" |
| ], |
| "force_upcast": true, |
| "in_channels": 3, |
| "latent_channels": 4, |
| "latents_mean": null, |
| "latents_std": null, |
| "layers_per_block": 2, |
| "mid_block_add_attention": true, |
| "norm_num_groups": 32, |
| "out_channels": 3, |
| "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" |
| }, |
| "sample_size": 512, |
| "scaling_factor": 0.18215, |
| "shift_factor": null, |
| "up_block_types": [ |
| "UpDecoderBlock2D", |
| "UpDecoderBlock2D", |
| "UpDecoderBlock2D", |
| "UpDecoderBlock2D" |
| ], |
| "use_post_quant_conv": true, |
| "use_quant_conv": true |
| } |
|
|
