diff --git a/scripts/evaluation/world_model_interaction.py b/scripts/evaluation/world_model_interaction.py
index 2f2d690..75f9c98 100644
--- a/scripts/evaluation/world_model_interaction.py
+++ b/scripts/evaluation/world_model_interaction.py
@@ -912,6 +912,10 @@ def run_inference(args: argparse.Namespace, gpu_num: int, gpu_no: int) -> None:
         diffusion_autocast_dtype = torch.bfloat16
         print(">>> diffusion backbone set to bfloat16")
 
+    if hasattr(model, "projector_bf16"):
+        model.projector_bf16 = args.projector_dtype == "bf16"
+        print(f">>> projector dtype set to {args.projector_dtype}")
+
     log_inference_precision(model)
 
     profiler.record_memory("after_model_load")
@@ -1255,6 +1259,13 @@ def get_parser():
         default="fp32",
         help="Dtype for diffusion backbone weights and sampling autocast."
     )
+    parser.add_argument(
+        "--projector_dtype",
+        type=str,
+        choices=["fp32", "bf16"],
+        default="fp32",
+        help="Dtype for image/state/action projectors (autocast in forward)."
+    )
     parser.add_argument(
         "--n_action_steps",
         type=int,
diff --git a/unitree_g1_pack_camera/case1/run_world_model_interaction.sh b/unitree_g1_pack_camera/case1/run_world_model_interaction.sh
index 3054dd3..a6dd008 100644
--- a/unitree_g1_pack_camera/case1/run_world_model_interaction.sh
+++ b/unitree_g1_pack_camera/case1/run_world_model_interaction.sh
@@ -21,5 +21,6 @@ dataset="unitree_g1_pack_camera"
         --timestep_spacing 'uniform_trailing' \
         --guidance_rescale 0.7 \
         --perframe_ae \
-        --diffusion_dtype bf16
+        --diffusion_dtype bf16 \
+        --projector_dtype bf16
 } 2>&1 | tee "${res_dir}/output.log"
diff --git a/useful.sh b/useful.sh
index aec7681..bcbf234 100644
--- a/useful.sh
+++ b/useful.sh
@@ -71,3 +71,11 @@ python3 psnr_score_for_challenge.py --gt_video unitree_g1_pack_camera/case1/unit
   - 优先 BF16：稳定性好于 FP16
   - 只做半精度，不做 INT 量化：保持 PSNR
   - VAE 尽量 FP32：最影响画质的模块
+
+BF16 projector比FP32 projector更准的可能原因：
+  - 数值路径更一致：主干在 BF16 下做 attention/MLP，projector 若是 FP32 会在进入主干前被 downcast，导致“先高精度非线性→再截断”的分布偏移；直接 BF16 算 projector 反而让输出
+    分布更贴近主干的计算习惯。
+  - 训练分布匹配：训练时你是 precision:16，projector 长期在低精度环境下被优化；推理用 FP32 反而可能偏离训练时的统计特性。
+  - LayerNorm/Softmax 敏感：Resampler/MLP 里 LN/Softmax 对精度很敏感，FP32 计算后再降精度，数值边界更容易“硬截断”；BF16 全程计算可能更平滑。
+
+  这也解释了为什么你看到 BF16 projector 反而更准。
\ No newline at end of file