chore: add LICENSE, CITATION.cff, .env.example, expand .gitignore

.env.example

@@ -0,0 +1,11 @@
+# Hugging Face Hub (write scope) — used by scripts/download_checkpoint.py and
+# any local fine-tune resume helpers that pull/push from atandra2000/sd-from-scratch-v1.
+HF_TOKEN=
+
+# Weights & Biases — used by SD_Train.py for training-time monitoring.
+WANDB_API_KEY=
+
+# Optional: override paths if you don't use the defaults
+# CHECKPOINT_DIR=./checkpoints
+# LATENT_DIR=./laion_latents
+# HF_HOME=./.hf-cache

.gitignore

@@ -1,18 +1,37 @@
+# Python
 __pycache__/
 *.pyc
 *.pyo
+*.pyd
 .Python
 *.egg-info/
 dist/
 build/
+.pytest_cache/
+.ruff_cache/
+.mypy_cache/
+.coverage
+htmlcov/
+
+# Virtual environments
 .env
+.env.local
+.env.*.local
 .venv
 venv/
+diffusion/
+
+# Secrets (defensive — never commit raw token files)
+secrets/
+*.pem
+*.key
 
 # Checkpoints and model weights
 *.pt
 *.pth
+*.ckpt
 *.safetensors
+*.bin
 checkpoints/
 outputs/
 
@@ -21,27 +40,50 @@ outputs/
 *.parquet
 *.tar
 *.zip
+*.tgz
+*.h5
+*.hdf5
+*.arrow
+*.lance
 laion_*/
+vggface_*/
+coco_*/
+dm_*/
+p7_*/
 /workspace/
 
-# Training logs and wandb
+# HF/HuggingFace caches
+.hf-cache/
+.cache/huggingface/
+
+# Training logs and W&B
 *.log
+logs/
 wandb/
 training.log
+runs/
+tb_logs/
 
 # Generated images (exclude training validation grids, keep curated samples)
 val_epoch_*.png
 results/*.png
 !results/samples/
 !results/samples/*.png
 !assets/*.png
+!docs/images/
 
 # Jupyter
 .ipynb_checkpoints/
 *.ipynb
 
-# IDE
+# IDE / editor
 .idea/
 .vscode/
 *.swp
+*.swo
+*~
+
+# OS
 .DS_Store
+Thumbs.db
+desktop.ini

CITATION.cff

@@ -0,0 +1,39 @@
+cff-version: 1.2.0
+title: "Stable Diffusion from Scratch (SD-From-Scratch v1)"
+message: "If you use this software or the released checkpoint, please cite it as below."
+type: software
+authors:
+  - given-names: Atandra
+    family-names: Bharati
+    email: atandrabharati@gmail.com
+repository-code: "https://github.com/atandra2000/StableDiffusion"
+url: "https://github.com/atandra2000/StableDiffusion"
+abstract: >-
+  A Stable Diffusion 1.x-class latent diffusion model implemented and
+  trained entirely from scratch in PyTorch on 2× RTX 5090 (Blackwell) GPUs.
+  Provides the full UNet (~860M params), DDPM/DDIM schedulers, LAION data
+  pipeline, dual-GPU latent encoding, DDP+BF16 training loop, and
+  CUDA/MPS inference. Released checkpoint sd_epoch_042.pt is hosted on
+  Hugging Face Hub at atandra2000/sd-from-scratch-v1.
+license: MIT
+version: "1.0.0"
+date-released: 2026-06-02
+keywords:
+  - stable-diffusion
+  - latent-diffusion
+  - diffusion-models
+  - pytorch
+  - text-to-image
+  - generative-ai
+  - from-scratch
+  - rtx-5090
+  - blackwell
+  - ddp
+preferred-citation:
+  type: software
+  authors:
+    - given-names: Atandra
+      family-names: Bharati
+  title: "SD-From-Scratch v1: A Stable-Diffusion-class latent diffusion model trained from scratch on dual RTX 5090s"
+  year: 2026
+  url: "https://huggingface.co/atandra2000/sd-from-scratch-v1"

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Atandra Bharati
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.