---
language: en
library_name: pytorch
tags:
  - pytorch
  - gan
  - week06-caa
  - conditional-gan
---

# mnist-conditional-gan

**Local run name:** `gan_run`

Generator weights exported from the Week 6 GAN lab training codebase.

## Repository files

| Key | Value |
| --- | --- |
| `mnist_cgan_generator.pth` | Generator `state_dict` (PyTorch) |
| `generator_training_config.json` | Training hyperparameters / layout (JSON) |
| `README.md` | This model card (auto-generated on upload) |

## Export provenance

| Key | Value |
| --- | --- |
| Checkpoint path | `/home/ucloud/caa-andre/complements-ml-labs-andreribeiro87/week-06/outputs/runs/gan_run__20260322T142324Z__9e0ad9f4/checkpoints/latest.pt` |
| Run directory JSON | `config.json`, `run_meta.json` |

## Run metadata (`run_meta.json`)

| Key | Value |
| --- | --- |
| config_fingerprint | 9e0ad9f4 |
| created_utc | 20260322T142324Z |
| run_name | gan_run |

## Training configuration (summary)

| Key | Value |
| --- | --- |
| run_name | gan_run |
| task | mnist |
| image_size | 64 |
| nz | 100 |
| ngf | 64 |
| ndf | 64 |
| num_classes | 10 |
| adversarial_loss | hinge |
| use_gradient_penalty | False |
| lambda_fm | 0.0 |
| lambda_perc | 0.0 |
| epochs | 25 |
| max_steps | None |
| batch_size | 32 |
| lr_g | 0.0002 |
| lr_d | 0.0002 |
| n_critic | 1 |
| seed | 42 |
| spectral_norm_d | True |

## Source JSON (run directory)

Verbatim JSON files from the run folder root (next to `checkpoints/`).

### `config.json`

```json
{
  "adversarial_loss": "hinge",
  "batch_size": 32,
  "beta1": 0.5,
  "beta2": 0.999,
  "checkpoint_every": 1000,
  "data_cache_dir": null,
  "device": "cuda",
  "epochs": 25,
  "eval_at_end": true,
  "eval_batch_size": 32,
  "eval_every": 1000,
  "eval_num_fake": 2048,
  "eval_num_real": 2048,
  "image_size": 64,
  "lambda_fm": 0.0,
  "lambda_gp": 10.0,
  "lambda_perc": 0.0,
  "log_every": 50,
  "lr_d": 0.0002,
  "lr_g": 0.0002,
  "max_steps": null,
  "metrics_device": null,
  "n_critic": 1,
  "n_samples_grid": 64,
  "ndf": 64,
  "ngf": 64,
  "num_classes": 10,
  "num_workers": 2,
  "nz": 100,
  "output_root": "outputs",
  "run_name": "gan_run",
  "sample_every": 500,
  "seed": 42,
  "spectral_norm_d": true,
  "tags": {},
  "task": "mnist",
  "use_bf16": true,
  "use_gradient_penalty": false,
  "use_wandb": true,
  "wandb_entity": null,
  "wandb_mode": "online",
  "wandb_project": "week06-gan"
}
```

### `run_meta.json`

```json
{
  "config": {
    "adversarial_loss": "hinge",
    "batch_size": 32,
    "beta1": 0.5,
    "beta2": 0.999,
    "checkpoint_every": 1000,
    "data_cache_dir": null,
    "device": "cuda",
    "epochs": 25,
    "eval_at_end": true,
    "eval_batch_size": 32,
    "eval_every": 1000,
    "eval_num_fake": 2048,
    "eval_num_real": 2048,
    "image_size": 64,
    "lambda_fm": 0.0,
    "lambda_gp": 10.0,
    "lambda_perc": 0.0,
    "log_every": 50,
    "lr_d": 0.0002,
    "lr_g": 0.0002,
    "max_steps": null,
    "metrics_device": null,
    "n_critic": 1,
    "n_samples_grid": 64,
    "ndf": 64,
    "ngf": 64,
    "num_classes": 10,
    "num_workers": 2,
    "nz": 100,
    "output_root": "outputs",
    "run_name": "gan_run",
    "sample_every": 500,
    "seed": 42,
    "spectral_norm_d": true,
    "tags": {},
    "task": "mnist",
    "use_bf16": true,
    "use_gradient_penalty": false,
    "use_wandb": true,
    "wandb_entity": null,
    "wandb_mode": "online",
    "wandb_project": "week06-gan"
  },
  "config_fingerprint": "9e0ad9f4",
  "created_utc": "20260322T142324Z",
  "run_name": "gan_run"
}
```