Release STCOcc-finetuned six-frame World Model checkpoint

README.md

@@ -9,28 +9,37 @@ tags:
 
 # LO-World
 
-Pretrained PyTorch checkpoints for UniAV-World, a BEV latent world model for joint future occupancy and LiDAR prediction.
+Pretrained PyTorch checkpoints for LO-World, a BEV latent world model for joint future occupancy and LiDAR prediction.
 
 ## Download
 
 ```bash
 hf download JimmyLiaoo/LO-World \
   bev_world_12f_autoregressive.pth bev_world_12f_layout_autoregressive.pth \
-  bev_world_6f_waymo.pth occ_lidar_vae.pth SHA256SUMS --local-dir ckpts
+  bev_world_6f_stc.pth bev_world_6f_waymo.pth occ_lidar_vae.pth SHA256SUMS --local-dir ckpts
 ```
 
 ## Checkpoints
 
-The release contains four checkpoints for inference and result reproduction.
+The release contains five checkpoints for inference and result reproduction.
 
 | File | Model |
 | --- | --- |
 | bev_world_12f_autoregressive.pth | 12-frame autoregressive World Model |
 | bev_world_12f_layout_autoregressive.pth | 12-frame autoregressive World Model with physical v1 BEVLayout conditioning |
+| bev_world_6f_stc.pth | 6-frame nuScenes World Model fine-tuned on STCOcc-conditioned latents |
 | bev_world_6f_waymo.pth | 6-frame Waymo World Model |
 | occ_lidar_vae.pth | Joint occupancy-LiDAR tokenizer |
 
-Use the corresponding configuration from the UniAV-World code repository. The Layout checkpoint is the physical v1 fine-tuned epoch 8 EMA and must be paired with the corrected layouts extracted to `data/bev_layout/`. World Model checkpoints include the tokenizer parameters. Training and evaluation require separately prepared dataset files and tokenizer latents.
+Use the corresponding configuration from the LO-World code repository. The Layout checkpoint is the physical v1 fine-tuned epoch 8 EMA and must be paired with the corrected layouts extracted to `data/bev_layout/`. World Model checkpoints include the tokenizer parameters. Training and evaluation require separately prepared dataset files and tokenizer latents.
+
+For `bev_world_6f_stc.pth`, use
+`configs/world_model/bev_world_6f_stc_occ.py` and matching STCOcc-conditioned
+validation latents at `data/stc_occ_lidar_latent/<sample_token>.npz`.
+This checkpoint contains its embedded tokenizer parameters, which differ from
+`occ_lidar_vae.pth`. Input latents must use the tokenizer employed during
+STCOcc fine-tuning. Evaluation targets remain the original Occ3D labels and
+LiDAR range images.
 
 ## Verify downloads
 

SHA256SUMS

@@ -1,4 +1,5 @@
 d38a6988f1da654ab7fb713e750aa8fe23fd4f208f15c18aca6a9812e161fbc9  bev_world_12f_autoregressive.pth
 a2c725491a70f34d6ac8962ba5b44bd2d5d8fb27e31aad787895d55e55fc4383  bev_world_12f_layout_autoregressive.pth
+b8f3f3d0552cf5aacc6b974e76f72c0dfac700e3008c69d450a39387d0fcdc63  bev_world_6f_stc.pth
 0662648dec4a5b13c4b3ef9fb96309adaaf578d0c667b772791c6617cbce495e  bev_world_6f_waymo.pth
 0ef9756b6e1df8e5e357b53421fb8a4be6fa76f0590332152ba6d58e39d3a901  occ_lidar_vae.pth

bev_world_6f_stc.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8f3f3d0552cf5aacc6b974e76f72c0dfac700e3008c69d450a39387d0fcdc63
+size 240367756