Document uploaded checkpoints

README.md

@@ -2,7 +2,7 @@
 license: mit
 ---
 
-# Diffusion Memorization — Model Checkpoints (placeholder)
+# Diffusion Memorization — Model Checkpoints
 
 Trained VAE and score-model checkpoints for the paper accompanying
 [trevorbchen/diffusion_memorization](https://github.com/trevorbchen/diffusion_memorization)
@@ -10,14 +10,47 @@ Trained VAE and score-model checkpoints for the paper accompanying
 [trevorbchen/diffusion_memroization](https://huggingface.co/datasets/trevorbchen/diffusion_memroization)
 (curated results / diverse-1k subset indices).
 
-> **Status: private, placeholder — no weights uploaded yet.** Accompanies a
-> paper currently under review. Please do not share this link or make it
-> public until the authors say otherwise.
+> **Status: private.** Accompanies a paper currently under review. Please do
+> not share this link or make it public until the authors say otherwise.
 
-**Scope:** this will hold checkpoints for the MLP-based results (synthetic
-RFNN/MLP, and real-data CelebA/CIFAR-10 VAE + MLP-score diffusion models).
-The spatial-DiT pilot experiment and its checkpoints are intentionally
-excluded from this release.
+**Scope:** these are the checkpoints behind the paper's headline MLP-based
+results. The spatial-DiT pilot experiment and its checkpoints are
+intentionally excluded from this release.
 
-Checkpoints pending upload — see the code repo for the exact training
-scripts and hyperparameters used to produce them.
+## Contents
+
+```
+vae_checkpoints/
+├── celeba_resnet_modernloss_d{10,15,20,...,200}/vae.pt   # final CelebA VAE, 14 widths
+└── cifar10_resnet_beta005_d{10,20,...,260}/vae.pt         # final CIFAR-10 VAE, 19 widths
+
+results/
+├── celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/
+│   └── seed{42..N}/d{10,...,200}/
+│       ├── last_model.pt    # MLP score-network checkpoint (hidden=1024, depth=5)
+│       ├── config.json      # exact training hyperparameters
+│       ├── metrics.jsonl    # per-eval-step memorization_fraction_pixel, FID, loss
+│       └── runner.log       # training log (includes any NaN/retry history)
+└── cifar10_beta005_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/
+    └── seed{42..N}/d{20,...,260}/   (same structure)
+```
+
+VAEs: ResNet conv architecture, beta-VAE with free bits and capacity warmup
+(`--arch resnet`, see the code repo's `standard_vae.py` /
+`train_vae_celeba_standard_tar.py` for exact args). Score models: SGD
+(lr=0.001, momentum=0.80), trained against the diverse-1k subset in the
+dataset repo, 5,000,000 steps, memorization/FID evaluated every 100k steps
+against 10k generated samples in pixel space.
+
+Memorization ratio: `d(gen, NN1_train) / d(gen, NN2_train)`, memorized if
+`< 1/3` (Bonnaire/Somepalli). See `lib/metrics.py` in the code repo.
+
+Some `d260` CIFAR-10 runs went non-finite under this fixed LR/momentum at
+high latent width — see `runner.log` for affected seeds; the paper's clean
+tables use replacement seeds for those cells rather than a different
+optimizer configuration (kept consistent across all `d` for a fair
+comparison).
+
+## License
+
+MIT (matching the code repository).