Upload folder using huggingface_hub

README.md

@@ -1,3 +1,111 @@
----
-license: mit
----
+# Flow Matching CIFAR-10 Model
+
+A flow matching model for unconditional image generation trained on the CIFAR-10 dataset. This model uses continuous normalizing flows with the FlowMatchEulerDiscreteScheduler for efficient sampling.
+
+## Model Details
+
+- **Architecture**: UNet2DModel with flow matching
+- **Dataset**: CIFAR-10 (32x32 RGB images)
+- **Scheduler**: FlowMatchEulerDiscreteScheduler
+- **Training Steps**: 1000 timesteps
+- **Framework**: Diffusers 0.35.0.dev0
+
+## Flow Matching Configuration
+
+The model uses FlowMatchEulerDiscreteScheduler with the following key parameters:
+- **Base shift**: 0.5
+- **Shift**: 1.0 (exponential time shifting)
+- **Base image sequence length**: 256
+- **Max image sequence length**: 4096
+- **Stochastic sampling**: Disabled for deterministic generation
+
+## Usage
+
+### Basic Generation
+
+```python
+from diffusers import DDPMPipeline
+
+# Load the flow matching model
+pipeline = DDPMPipeline.from_pretrained("FrankCCCCC/cfm-cifar10-32")
+
+# Generate an image
+image = pipeline().images[0]
+image.save("generated_cifar10.png")
+```
+
+### Custom Inference Steps
+
+```python
+from diffusers import DDPMPipeline
+
+pipeline = DDPMPipeline.from_pretrained("FrankCCCCC/cfm-cifar10-32")
+
+# Generate with custom number of inference steps
+num_inference_steps: int = 1000
+pipeline.scheduler.set_timesteps(num_inference_steps)
+image = pipeline().images[0]
+image.save("fast_generated_cifar10.png")
+```
+
+### Batch Generation
+
+```python
+from diffusers import DDPMPipeline
+
+pipeline = DDPMPipeline.from_pretrained("FrankCCCCC/cfm-cifar10-32")
+
+# Generate multiple images at once
+images = pipeline(batch_size=4).images
+for i, image in enumerate(images):
+    image.save(f"generated_cifar10_{i}.png")
+```
+
+## Flow Matching vs Standard Diffusion
+
+This model implements flow matching, which offers several advantages over standard diffusion models:
+- **Faster sampling**: More efficient ODE solving with fewer steps
+- **Better training stability**: Continuous normalizing flows provide smoother optimization
+- **Flexible scheduling**: Exponential time shifting for improved sample quality
+
+## Model Architecture
+
+- **UNet**: Standard UNet2DModel for denoising/flow prediction
+- **Scheduler**: FlowMatchEulerDiscreteScheduler with exponential time shifting
+- **Output**: 32x32 RGB images matching CIFAR-10 distribution
+
+## Requirements
+
+```bash
+pip install diffusers torch torchvision
+```
+
+## Samples
+1. ![sample_1](https://huggingface.co/FrankCCCCC/cfm-cifar10-32/resolve/main/images/sample0.jpg)
+2. ![sample_2](https://huggingface.co/FrankCCCCC/cfm-cifar10-32/resolve/main/images/sample1.jpg)
+3. ![sample_3](https://huggingface.co/FrankCCCCC/cfm-cifar10-32/resolve/main/images/sample2.jpg)
+4. ![sample_4](https://huggingface.co/FrankCCCCC/cfm-cifar10-32/resolve/main/images/sample3.jpg)
+5. ![sample_5](https://huggingface.co/FrankCCCCC/cfm-cifar10-32/resolve/main/images/sample4.jpg)
+
+## Citation
+
+If you use this model, please cite the original flow matching and diffusion literature:
+
+```bibtex
+@inproceedings{DDPM,
+  author = {Ho, Jonathan and Jain, Ajay and Abbeel, Pieter},
+  booktitle = {Advances in Neural Information Processing Systems},
+  title = {Denoising Diffusion Probabilistic Models},
+  url = {https://proceedings.neurips.cc/paper_files/paper/2020/file/4c5bcfec8584af0d967f1ab10179ca4b-Paper.pdf},
+  year = {2020}
+}
+
+@inproceedings{FM,
+  title={Flow Matching for Generative Modeling},
+  author={Yaron Lipman and Ricky T. Q. Chen and Heli Ben-Hamu and Maximilian Nickel and Matthew Le},
+  booktitle={The Eleventh International Conference on Learning Representations },
+  year={2023},
+  url={https://openreview.net/forum?id=PqvMRDCJT9t}
+}
+
+```

config.json

@@ -0,0 +1,29 @@
+{
+    "beta_1": 0.9,
+    "beta_2": 0.999,
+    "epsilon": 1e-08,
+    "lr_sched_num_warmup_steps": 45000,
+    "lr_sched_lr_end": 1e-07,
+    "lr_sched_power": 1.0,
+    "ep_model_dir": "epochs",
+    "output_dir": "fm_cifar10",
+    "ckpt_dir": "ckpt",
+    "data_ckpt_dir": "data.ckpt",
+    "is_save_all_model_epochs": false,
+    "args_key": "args",
+    "default_key": "default",
+    "final_key": "final",
+    "config_file": "config.json",
+    "project": "cfm-training",
+    "run_name": "train_cfm",
+    "model_id": "google/ddpm-cifar10-32",
+    "batch_size": 256,
+    "num_epochs": 1000,
+    "lr": 0.0005,
+    "weight_decay": 0.0,
+    "num_train_timesteps": 1000,
+    "num_inference_steps": 1000,
+    "sigma_min": 0.0,
+    "seed": 42,
+    "device": "cuda:0"
+}

model_index.json

@@ -0,0 +1,13 @@
+{
+  "_class_name": "DDPMPipeline",
+  "_diffusers_version": "0.35.0.dev0",
+  "_name_or_path": "/home/sc3379/workspace/research/cfm-cifar10-32",
+  "scheduler": [
+    "diffusers",
+    "FlowMatchEulerDiscreteScheduler"
+  ],
+  "unet": [
+    "diffusers",
+    "UNet2DModel"
+  ]
+}

scheduler/scheduler_config.json

@@ -0,0 +1,18 @@
+{
+  "_class_name": "FlowMatchEulerDiscreteScheduler",
+  "_diffusers_version": "0.35.0.dev0",
+  "base_image_seq_len": 256,
+  "base_shift": 0.5,
+  "invert_sigmas": false,
+  "max_image_seq_len": 4096,
+  "max_shift": 1.15,
+  "num_train_timesteps": 1000,
+  "shift": 1.0,
+  "shift_terminal": null,
+  "stochastic_sampling": false,
+  "time_shift_type": "exponential",
+  "use_beta_sigmas": false,
+  "use_dynamic_shifting": false,
+  "use_exponential_sigmas": false,
+  "use_karras_sigmas": false
+}

unet/config.json

@@ -0,0 +1,48 @@
+{
+  "_class_name": "UNet2DModel",
+  "_diffusers_version": "0.35.0.dev0",
+  "_name_or_path": "/home/sc3379/workspace/research/cfm-cifar10-32/unet",
+  "act_fn": "silu",
+  "add_attention": true,
+  "attention_head_dim": null,
+  "attn_norm_num_groups": null,
+  "block_out_channels": [
+    128,
+    256,
+    256,
+    256
+  ],
+  "center_input_sample": false,
+  "class_embed_type": null,
+  "down_block_types": [
+    "DownBlock2D",
+    "AttnDownBlock2D",
+    "DownBlock2D",
+    "DownBlock2D"
+  ],
+  "downsample_padding": 0,
+  "downsample_type": "conv",
+  "dropout": 0.0,
+  "flip_sin_to_cos": false,
+  "freq_shift": 1,
+  "in_channels": 3,
+  "layers_per_block": 2,
+  "mid_block_scale_factor": 1,
+  "mid_block_type": "UNetMidBlock2D",
+  "norm_eps": 1e-06,
+  "norm_num_groups": 32,
+  "num_class_embeds": null,
+  "num_train_timesteps": null,
+  "out_channels": 3,
+  "resnet_time_scale_shift": "default",
+  "sample_size": 32,
+  "time_embedding_dim": null,
+  "time_embedding_type": "positional",
+  "up_block_types": [
+    "UpBlock2D",
+    "UpBlock2D",
+    "AttnUpBlock2D",
+    "UpBlock2D"
+  ],
+  "upsample_type": "conv"
+}

unet/diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97d25692dbd390a357e7375966ecd521418d7a9623a01037dc1aeef809142980
+size 143020060