Instructions to use exact-railcar/nvidia with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use exact-railcar/nvidia with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("exact-railcar/nvidia", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
Download config.json from exact-railcar/nvidia: direct link, hf CLI and curl.
- Browser
- Download file 3.22 kB
-
https://huggingface.co/exact-railcar/nvidia/resolve/main/config.json
- Command line
-
hf download hf://exact-railcar/nvidia/config.json
-
curl -L -o config.json https://huggingface.co/exact-railcar/nvidia/resolve/main/config.json
3.22 kB
| { | |
| "_class_name": "WanVACETransformer3DModel", | |
| "_diffusers_version": "0.41.0", | |
| "_name_or_path": "/tmp/kaggle", | |
| "added_kv_proj_dim": null, | |
| "attention_head_dim": 128, | |
| "cross_attn_norm": true, | |
| "eps": 1e-06, | |
| "ffn_dim": 13824, | |
| "freq_dim": 256, | |
| "image_dim": null, | |
| "in_channels": 16, | |
| "num_attention_heads": 40, | |
| "num_layers": 40, | |
| "out_channels": 16, | |
| "patch_size": [ | |
| 1, | |
| 2, | |
| 2 | |
| ], | |
| "pos_embed_seq_len": null, | |
| "qk_norm": "rms_norm_across_heads", | |
| "quantization_config": { | |
| "block_quantize": null, | |
| "calib_cfg": { | |
| "method": "max" | |
| }, | |
| "channel_quantize": null, | |
| "disable_conv_quantization": false, | |
| "forward_loop": null, | |
| "modelopt_config": { | |
| "algorithm": { | |
| "method": "max" | |
| }, | |
| "quant_cfg": { | |
| "*block_sparse_moe.gate*": { | |
| "enable": false | |
| }, | |
| "*input_quantizer": { | |
| "enable": false | |
| }, | |
| "*k_bmm_quantizer": { | |
| "enable": false | |
| }, | |
| "*linear_attn.conv1d*": { | |
| "enable": false | |
| }, | |
| "*lm_head*": { | |
| "enable": false | |
| }, | |
| "*mixer.conv1d*": { | |
| "enable": false | |
| }, | |
| "*mlp.gate.*": { | |
| "enable": false | |
| }, | |
| "*mlp.shared_expert_gate.*": { | |
| "enable": false | |
| }, | |
| "*norm1*": { | |
| "enable": false | |
| }, | |
| "*norm2*": { | |
| "enable": false | |
| }, | |
| "*norm3*": { | |
| "enable": false | |
| }, | |
| "*output_layer*": { | |
| "enable": false | |
| }, | |
| "*output_quantizer": { | |
| "enable": false | |
| }, | |
| "*proj_out.*": { | |
| "enable": false | |
| }, | |
| "*q_bmm_quantizer": { | |
| "enable": false | |
| }, | |
| "*rope*": { | |
| "enable": false | |
| }, | |
| "*router*": { | |
| "enable": false | |
| }, | |
| "*scale_shift_table*": { | |
| "enable": false | |
| }, | |
| "*softmax_quantizer": { | |
| "enable": false | |
| }, | |
| "*time_embedder*": { | |
| "enable": false | |
| }, | |
| "*v_bmm_quantizer": { | |
| "enable": false | |
| }, | |
| "*weight_quantizer": { | |
| "fake_quant": false, | |
| "num_bits": 4 | |
| }, | |
| "default": { | |
| "enable": false | |
| }, | |
| "nn.BatchNorm1d": { | |
| "*": { | |
| "enable": false | |
| } | |
| }, | |
| "nn.BatchNorm2d": { | |
| "*": { | |
| "enable": false | |
| } | |
| }, | |
| "nn.BatchNorm3d": { | |
| "*": { | |
| "enable": false | |
| } | |
| }, | |
| "nn.LeakyReLU": { | |
| "*": { | |
| "enable": false | |
| } | |
| }, | |
| "output.*": { | |
| "enable": false | |
| } | |
| } | |
| }, | |
| "modules_to_not_convert": [ | |
| "rope", | |
| "time_embedder", | |
| "scale_shift_table", | |
| "norm1", | |
| "norm2", | |
| "norm3" | |
| ], | |
| "quant_method": "modelopt", | |
| "quant_type": "INT4", | |
| "scale_block_quantize": null, | |
| "scale_channel_quantize": null, | |
| "weight_only": true | |
| }, | |
| "rope_max_seq_len": 1024, | |
| "text_dim": 4096, | |
| "vace_in_channels": 96, | |
| "vace_layers": [ | |
| 0, | |
| 5, | |
| 10, | |
| 15, | |
| 20, | |
| 25, | |
| 30, | |
| 35 | |
| ] | |
| } | |