Image-to-Image
Cosmos
PyTorch
nvidia
cosmos-predict2
diffusion
inpainting
anomaly-generation
synthetic-data-generation
pcb-inspection
few-shot
fine-tuned
Instructions to use nvidia/Cosmos-AnomalyGen-PCB-2B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Cosmos
How to use nvidia/Cosmos-AnomalyGen-PCB-2B with Cosmos:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
| # DDP config template (default, use with predict2_anomaly_gen_ddp_{2b,14b}). | |
| # Placeholders in <angle-brackets> are filled by scripts/generate_config.py. | |
| # Keep sections minimal — the experiment provides defaults for everything else. | |
| job: | |
| project: anomaly_gen | |
| group: UC1_data_byTexture_rot90aug_newForTest | |
| name: UC1_data_byTexture_rot90aug_newForTest_training_FP32_lr0.02_bs=2_2B_512x512 | |
| optimizer: | |
| lr: 0.02 | |
| checkpoint: | |
| save_iter: 2000 | |
| trainer: | |
| max_iter: 28000 | |
| logging_iter: 10 | |
| validation_iter: 2000 | |
| run_validation: True | |
| early_stop: | |
| enabled: false | |
| metric: nn | |
| patience: 5 | |
| min_delta: 0 | |
| min_delta_mode: rel | |
| dataloader_train: | |
| batch_size: 2 | |
| dataset: | |
| dataset_dir: /workspace/cosmos-anomalygen-predict2/datasets/CPTX_dryrun/Dataset/UC1_data_byTexture_rot90aug_newForTest | |
| image_size: | |
| - 512 | |
| - 512 | |
| anomaly_types: [[IC, bridge], [passive_component, excess_solder], [passive_component, missing]] | |
| seed: 1 | |
| data_augprob: 0.5 | |
| aug_type: random_ratio_crop | |
| ratio_range: [1.5, 8.0] | |
| dataloader_val: | |
| batch_size: 32 | |
| dataset: | |
| input_data_path: ag_inference/validation_UC1_data_byTexture_rot90aug_newForTest/testcase.jsonl | |
| model: | |
| config: | |
| ag_config: | |
| ad_precision: float32 | |
| t5_model_name: checkpoints/google-t5/t5-large | |
| anomaly_embedding: | |
| anomaly_types: [[IC, bridge], [passive_component, excess_solder], [passive_component, missing]] | |
| freeze: False | |
| mask_encoder: | |
| encoder_config: | |
| init_cfg: | |
| checkpoint: checkpoints/NVDINOV2/nv_dinov2_classification_model.ckpt | |