olivame/unifolm-world-model-action
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

- state_unet 放到一个独立的 CUDA stream 上执行

Browse Source 
- action_unet 在默认 stream 上同时执行
  - 用 wait_stream 确保两者都完成后再返回
两个 1D UNet 输入完全独立，共享的 hs_a 和 context_action 都是只读的。GPU 利用率只有 ~31%，小张量 kernel 不会打满 GPU，两个 stream 可以真正并行。
This commit is contained in:
olivame
2026-02-10 10:47:10 +00:00
parent db848bca01
commit a09d35ae5b
3 changed files with 22 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/unifolm_wma/modules/networks/wma_model.py
View File

@@ -848,15 +848,16 @@ class WMAModel(nn.Module):
if not self.base_model_gen_only:
ba, _, _ = x_action.shape
ts_state = timesteps[:ba] if b > 1 else timesteps
# Run action_unet and state_unet in parallel via CUDA streams
s_stream = torch.cuda.Stream()
s_stream.wait_stream(torch.cuda.current_stream())
with torch.cuda.stream(s_stream):
s_y = self.state_unet(x_state, ts_state, hs_a,
context_action[:2], **kwargs)
a_y = self.action_unet(x_action, timesteps[:ba], hs_a,
context_action[:2], **kwargs)
# Predict state
if b > 1:
s_y = self.state_unet(x_state, timesteps[:ba], hs_a,
context_action[:2], **kwargs)
else:
s_y = self.state_unet(x_state, timesteps, hs_a,
context_action[:2], **kwargs)
torch.cuda.current_stream().wait_stream(s_stream)
else:
a_y = torch.zeros_like(x_action)
s_y = torch.zeros_like(x_state)

26
unitree_z1_dual_arm_cleanup_pencils/case1/output.log
View File

@@ -1,14 +1,14 @@
/mnt/ASC1637/miniconda3/envs/unifolm-wma-o/lib/python3.10/site-packages/lightning_fabric/__init__.py:29: UserWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html. The pkg_resources package is slated for removal as early as 2025-11-30. Refrain from using this package or pin to Setuptools<81.
__import__("pkg_resources").declare_namespace(__name__)
2026-02-10 06:42:14.444321: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
2026-02-10 06:42:14.447338: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.
2026-02-10 06:42:14.478442: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2026-02-10 06:42:14.478474: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2026-02-10 06:42:14.480279: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2026-02-10 06:42:14.488343: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.
2026-02-10 06:42:14.488598: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
2026-02-10 10:36:44.797852: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
2026-02-10 10:36:44.801300: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.
2026-02-10 10:36:44.837891: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2026-02-10 10:36:44.837946: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2026-02-10 10:36:44.839880: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2026-02-10 10:36:44.849073: I external/local_tsl/tsl/cuda/cudart_stub.cc:31] Could not find cuda drivers on your machine, GPU will not be used.
2026-02-10 10:36:44.849365: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX512F AVX512_VNNI AVX512_BF16 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2026-02-10 06:42:15.109100: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
2026-02-10 10:36:45.644793: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
[rank: 0] Global seed set to 123
/mnt/ASC1637/miniconda3/envs/unifolm-wma-o/lib/python3.10/site-packages/kornia/feature/lightglue.py:44: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
@torch.cuda.amp.custom_fwd(cast_inputs=torch.float32)
@@ -25,7 +25,7 @@ INFO:root:Loading pretrained ViT-H-14 weights (laion2b_s32b_b79k).
INFO:root:Loaded ViT-H-14 model config.
DEBUG:urllib3.connectionpool:https://hf-mirror.com:443 "HEAD /laion/CLIP-ViT-H-14-laion2B-s32B-b79K/resolve/main/open_clip_pytorch_model.bin HTTP/1.1" 302 0
INFO:root:Loading pretrained ViT-H-14 weights (laion2b_s32b_b79k).
/mnt/ASC1637/unifolm-world-model-action/scripts/evaluation/world_model_interaction.py:199: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
/mnt/ASC1637/unifolm-world-model-action/scripts/evaluation/world_model_interaction.py:198: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
state_dict = torch.load(ckpt, map_location="cpu")
>>> model checkpoint loaded.
>>> Load pre-trained model ...
@@ -116,7 +116,7 @@ DEBUG:PIL.Image:Importing WmfImagePlugin
DEBUG:PIL.Image:Importing WmfImagePlugin
DEBUG:PIL.Image:Importing XbmImagePlugin
DEBUG:PIL.Image:Importing XpmImagePlugin
DEBUG:PIL.Image:Importing XVThumbImagePlugin
DEBUG:PIL.Image:Importing XVThumbImagePlugin
12%|█▎ | 1/8 [01:06<07:46, 66.62s/it]
25%|██▌ | 2/8 [02:07<06:17, 62.97s/it]
@@ -140,6 +140,6 @@ DEBUG:PIL.Image:Importing XVThumbImagePlugin
>>> Step 4: generating actions ...
>>> Step 4: interacting with world model ...
>>>>>>>>>>>>>>>>>>>>>>>>
>>> Step 5: generating actions ...
>>> Step 5: interacting with world model ...
>>>>>>>>>>>>>>>>>>>>>>>>
>>> Step 5: generating actions ...
>>> Step 5: interacting with world model ...
>>>>>>>>>>>>>>>>>>>>>>>>

3
unitree_z1_dual_arm_cleanup_pencils/case1/run_world_model_interaction.sh
View File

@@ -22,6 +22,5 @@ dataset="unitree_z1_dual_arm_cleanup_pencils"
--guidance_rescale 0.7 \
--perframe_ae \
--vae_dtype bf16 \
--fast_policy_no_decode \
--sampler_type unipc
--fast_policy_no_decode
} 2>&1 | tee "${res_dir}/output.log"