Add stage 1/2/3 checkpoints, config.json, and model card

README.md

@@ -1,3 +1,42 @@
 ---
 license: apache-2.0
+datasets:
+- KlingTeam/MultiCamVideo-Dataset
+- KlingTeam/SynCamVideo-Dataset
+base_model:
+- alibaba-pai/Wan2.1-Fun-V1.1-1.3B-Control-Camera
+base_model_relation: finetune
+pipeline_tag: video-to-video
+tags:
+- pytorch
 ---
+
+This model was presented in [FlashRender: Few-Step Generative Rendering via Camera-Controlled Video MeanFlow](https://huggingface.co/papers/2609.03563).
+
+Please refer to the [Github](https://github.com/byeongjun-park/FlashRender) README for usage.
+
+## Checkpoints
+
+| Stage | File | Sampling |
+|:---:|---|---|
+| 1 | `epoch_20.safetensors` | 50 steps, `inference_mode=mul` |
+| 2 | `meanflow_epoch_20.safetensors` | 4 steps, `inference_mode=any` |
+| 3 | `onpolicy_epoch_5.safetensors` | 4 steps, `inference_mode=any` |
+
+**`onpolicy_epoch_5.safetensors` is the final model** and the default in `configs/base.yaml`.
+
+Each file stores only the **trainable parameter subset**, not a stand-alone diffusers/transformers
+checkpoint. Weights are `bfloat16`, except the relative-pose tokens and the per-block RoPE phase MLPs
+(`rel_pose_{src,tgt}_token`, `self_attn.rope_phase_{qk,vo}`), which are stored in `float32`.
+They are loaded with `strict=False` onto the base Wan DiT after it is patched by
+`flashrender_utils.model_utils.adjust_to_FlashRender`, so use the code repository:
+
+```python
+from huggingface_hub import snapshot_download
+
+snapshot_download(
+    repo_id="byeongjun-park/FlashRender",
+    local_dir="models/checkpoints",
+    allow_patterns=["*.safetensors", "config.json"],
+)
+```

config.json

@@ -0,0 +1,33 @@
+{
+  "model_name": "FlashRender",
+  "description": "Few-step generative rendering via camera-controlled video MeanFlow.",
+  "task": "video-to-video",
+  "license": "apache-2.0",
+  "code_repository": "https://github.com/byeongjun-park/FlashRender",
+  "base_model": "alibaba-pai/Wan2.1-Fun-V1.1-1.3B-Control-Camera",
+  "torch_dtype": "bfloat16",
+  "height": 480,
+  "width": 832,
+  "num_frames": 81,
+  "state_dict_format": "partial (trainable subset only); load with strict=False onto the base Wan DiT patched by flashrender_utils.model_utils.adjust_to_FlashRender",
+  "checkpoints": {
+    "epoch_20.safetensors": {
+      "stage": 1,
+      "training": "multi-step flow matching (t = r) with RETA feature alignment",
+      "num_inference_steps": 50,
+      "inference_mode": "mul"
+    },
+    "meanflow_epoch_20.safetensors": {
+      "stage": 2,
+      "training": "few-step MeanFlow distillation, resumed from stage 1",
+      "num_inference_steps": 4,
+      "inference_mode": "any"
+    },
+    "onpolicy_epoch_5.safetensors": {
+      "stage": 3,
+      "training": "on-policy flow-map distillation (DMD/VSD + GAN) of the stage-2 student",
+      "num_inference_steps": 4,
+      "inference_mode": "any"
+    }
+  }
+}

epoch_20.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5b6bf14b6e45eb74cc3a29c47e52e7b1e2f71faf84e08ba0f9b8ab8657f58f8
+size 1543319280

meanflow_epoch_20.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17928eb068db1a118d00d0c82275c8d1094813250ff7133b7df213ba9efa8168
+size 1582672592

onpolicy_epoch_5.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6152ef1bc25fc664674273e628e5a21dbdd9fe9f7a77264c865c3ade8e100a7c
+size 1657823568