evaluating-cosmos-policy

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Cosmos Policy Evaluation

Cosmos Policy 评估

Evaluation workflows for NVIDIA Cosmos Policy on LIBERO and RoboCasa simulation environments from the public

cosmos-policy

repository. Covers blank-machine setup, headless GPU evaluation, and inference profiling.

基于公开

cosmos-policy

仓库，在LIBERO和RoboCasa仿真环境中运行NVIDIA Cosmos Policy的评估工作流。内容覆盖裸机环境搭建、无桌面GPU评估和推理性能分析。

Quick start

快速开始

Run a minimal LIBERO evaluation using the official public eval module:

bash

uv run --extra cu128 --group libero --python 3.10 \
  python -m cosmos_policy.experiments.robot.libero.run_libero_eval \
    --config cosmos_predict2_2b_480p_libero__inference_only \
    --ckpt_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B \
    --config_file cosmos_policy/config/config.py \
    --use_wrist_image True \
    --use_proprio True \
    --normalize_proprio True \
    --unnormalize_actions True \
    --dataset_stats_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_dataset_statistics.json \
    --t5_text_embeddings_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_t5_embeddings.pkl \
    --trained_with_image_aug True \
    --chunk_size 16 \
    --num_open_loop_steps 16 \
    --task_suite_name libero_10 \
    --num_trials_per_task 1 \
    --local_log_dir cosmos_policy/experiments/robot/libero/logs/ \
    --seed 195 \
    --randomize_seed False \
    --deterministic True \
    --run_id_note smoke \
    --ar_future_prediction False \
    --ar_value_prediction False \
    --use_jpeg_compression True \
    --flip_images True \
    --num_denoising_steps_action 5 \
    --num_denoising_steps_future_state 1 \
    --num_denoising_steps_value 1 \
    --data_collection False

使用官方公开评估模块运行最简LIBERO评估：

bash

uv run --extra cu128 --group libero --python 3.10 \
  python -m cosmos_policy.experiments.robot.libero.run_libero_eval \
    --config cosmos_predict2_2b_480p_libero__inference_only \
    --ckpt_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B \
    --config_file cosmos_policy/config/config.py \
    --use_wrist_image True \
    --use_proprio True \
    --normalize_proprio True \
    --unnormalize_actions True \
    --dataset_stats_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_dataset_statistics.json \
    --t5_text_embeddings_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_t5_embeddings.pkl \
    --trained_with_image_aug True \
    --chunk_size 16 \
    --num_open_loop_steps 16 \
    --task_suite_name libero_10 \
    --num_trials_per_task 1 \
    --local_log_dir cosmos_policy/experiments/robot/libero/logs/ \
    --seed 195 \
    --randomize_seed False \
    --deterministic True \
    --run_id_note smoke \
    --ar_future_prediction False \
    --ar_value_prediction False \
    --use_jpeg_compression True \
    --flip_images True \
    --num_denoising_steps_action 5 \
    --num_denoising_steps_future_state 1 \
    --num_denoising_steps_value 1 \
    --data_collection False

Core concepts

核心概念

What Cosmos Policy is: NVIDIA Cosmos Policy is a vision-language-action (VLA) model that uses Cosmos Tokenizer to encode visual observations into discrete tokens, then predicts robot actions conditioned on language instructions and visual context.

Key architecture choices:

Component	Design
Visual encoder	Cosmos Tokenizer (discrete tokens)
Language conditioning	Cross-attention to language embeddings
Action prediction	Autoregressive action token generation

Public command surface: The supported evaluation entrypoints are

cosmos_policy.experiments.robot.libero.run_libero_eval

and

cosmos_policy.experiments.robot.robocasa.run_robocasa_eval

. Keep reproduction notes anchored to these public modules and their documented flags.

Cosmos Policy简介：NVIDIA Cosmos Policy是一款视觉-语言-动作（VLA）模型，它使用Cosmos Tokenizer将视觉观测编码为离散令牌，然后基于语言指令和视觉上下文预测机器人动作。

核心架构选型：

组件	设计
视觉编码器	Cosmos Tokenizer (离散令牌)
语言条件注入	对语言嵌入做交叉注意力
动作预测	自回归动作令牌生成

公开命令入口：支持的评估入口为

cosmos_policy.experiments.robot.libero.run_libero_eval

和

cosmos_policy.experiments.robot.robocasa.run_robocasa_eval

。复现说明请锚定这些公开模块及其文档化的参数。

Compute requirements

算力要求

Task	GPU	VRAM	Typical wall time
LIBERO smoke eval (1 trial)	1x A40/A100	~16 GB	5-10 min
LIBERO full eval (50 trials)	1x A40/A100	~16 GB	2-4 hours
RoboCasa single-task (2 trials)	1x A40/A100	~18 GB	10-15 min
RoboCasa all-tasks	1x A40/A100	~18 GB	4-8 hours

任务	GPU	VRAM	典型耗时
LIBERO冒烟评估（1次试跑）	1x A40/A100	~16 GB	5-10 分钟
LIBERO全量评估（50次试跑）	1x A40/A100	~16 GB	2-4 小时
RoboCasa单任务评估（2次试跑）	1x A40/A100	~18 GB	10-15 分钟
RoboCasa全任务评估	1x A40/A100	~18 GB	4-8 小时

When to use vs alternatives

适用场景与替代方案

Use this skill when:

Evaluating NVIDIA Cosmos Policy on LIBERO or RoboCasa benchmarks
Profiling inference latency and throughput for Cosmos Policy
Setting up headless EGL rendering for robot simulation on GPU clusters

Use alternatives when:

Training or fine-tuning Cosmos Policy from scratch (use official Cosmos training docs)
Working with OpenVLA-based policies (use
```
fine-tuning-openvla-oft
```
)
Working with Physical Intelligence pi0 models (use
```
fine-tuning-serving-openpi
```
)
Running real-robot evaluation rather than simulation

适用场景：

在LIBERO或RoboCasa基准上评估NVIDIA Cosmos Policy
分析Cosmos Policy的推理延迟和吞吐量
在GPU集群上为机器人仿真搭建无桌面EGL渲染环境

请使用替代方案的场景：

从头训练或微调Cosmos Policy（请使用官方Cosmos训练文档）
使用基于OpenVLA的策略（请使用
```
fine-tuning-openvla-oft
```
）
使用Physical Intelligence pi0模型（请使用
```
fine-tuning-serving-openpi
```
）
运行真机机器人评估而非仿真评估

Workflow 1: LIBERO evaluation

工作流1：LIBERO评估

Copy this checklist and track progress:

text

LIBERO Eval Progress:
- [ ] Step 1: Install environment and dependencies
- [ ] Step 2: Configure headless EGL rendering
- [ ] Step 3: Run smoke evaluation
- [ ] Step 4: Validate outputs and parse results
- [ ] Step 5: Run full benchmark if smoke passes

Step 1: Install environment

bash

git clone https://github.com/NVlabs/cosmos-policy.git
cd cosmos-policy

复制此检查清单跟踪进度：

text

LIBERO评估进度:
- [ ] 步骤1：安装环境和依赖
- [ ] 步骤2：配置无桌面EGL渲染
- [ ] 步骤3：运行冒烟评估
- [ ] 步骤4：验证输出并解析结果
- [ ] 步骤5：冒烟测试通过后运行全量基准测试

步骤1：安装环境

bash

git clone https://github.com/NVlabs/cosmos-policy.git
cd cosmos-policy

Follow SETUP.md to build and enter the supported Docker container.

参考SETUP.md构建并进入支持的Docker容器

Then, inside the container:

然后在容器内执行：

uv sync --extra cu128 --group libero --python 3.10


**Step 2: Configure headless rendering**

```bash
export CUDA_VISIBLE_DEVICES=0
export MUJOCO_EGL_DEVICE_ID=0
export MUJOCO_GL=egl
export PYOPENGL_PLATFORM=egl

Step 3: Run smoke evaluation

bash

uv run --extra cu128 --group libero --python 3.10 \
  python -m cosmos_policy.experiments.robot.libero.run_libero_eval \
    --config cosmos_predict2_2b_480p_libero__inference_only \
    --ckpt_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B \
    --config_file cosmos_policy/config/config.py \
    --use_wrist_image True \
    --use_proprio True \
    --normalize_proprio True \
    --unnormalize_actions True \
    --dataset_stats_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_dataset_statistics.json \
    --t5_text_embeddings_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_t5_embeddings.pkl \
    --trained_with_image_aug True \
    --chunk_size 16 \
    --num_open_loop_steps 16 \
    --task_suite_name libero_10 \
    --num_trials_per_task 1 \
    --local_log_dir cosmos_policy/experiments/robot/libero/logs/ \
    --seed 195 \
    --randomize_seed False \
    --deterministic True \
    --run_id_note smoke \
    --ar_future_prediction False \
    --ar_value_prediction False \
    --use_jpeg_compression True \
    --flip_images True \
    --num_denoising_steps_action 5 \
    --num_denoising_steps_future_state 1 \
    --num_denoising_steps_value 1 \
    --data_collection False

Step 4: Validate and parse results

python

import json
import glob

uv sync --extra cu128 --group libero --python 3.10


**步骤2：配置无桌面渲染**

```bash
export CUDA_VISIBLE_DEVICES=0
export MUJOCO_EGL_DEVICE_ID=0
export MUJOCO_GL=egl
export PYOPENGL_PLATFORM=egl

步骤3：运行冒烟评估

bash

uv run --extra cu128 --group libero --python 3.10 \
  python -m cosmos_policy.experiments.robot.libero.run_libero_eval \
    --config cosmos_predict2_2b_480p_libero__inference_only \
    --ckpt_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B \
    --config_file cosmos_policy/config/config.py \
    --use_wrist_image True \
    --use_proprio True \
    --normalize_proprio True \
    --unnormalize_actions True \
    --dataset_stats_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_dataset_statistics.json \
    --t5_text_embeddings_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_t5_embeddings.pkl \
    --trained_with_image_aug True \
    --chunk_size 16 \
    --num_open_loop_steps 16 \
    --task_suite_name libero_10 \
    --num_trials_per_task 1 \
    --local_log_dir cosmos_policy/experiments/robot/libero/logs/ \
    --seed 195 \
    --randomize_seed False \
    --deterministic True \
    --run_id_note smoke \
    --ar_future_prediction False \
    --ar_value_prediction False \
    --use_jpeg_compression True \
    --flip_images True \
    --num_denoising_steps_action 5 \
    --num_denoising_steps_future_state 1 \
    --num_denoising_steps_value 1 \
    --data_collection False

步骤4：验证并解析结果

python

import json
import glob

Find latest evaluation result from the official log directory

从官方日志目录查找最新的评估结果

log_files = sorted(glob.glob("cosmos_policy/experiments/robot/libero/logs/**/*.json", recursive=True)) with open(log_files[-1]) as f: results = json.load(f)

print(results)


**Step 5: Scale up**

Run across all four LIBERO task suites with 50 trials:

```bash
for suite in libero_spatial libero_object libero_goal libero_10; do
  uv run --extra cu128 --group libero --python 3.10 \
    python -m cosmos_policy.experiments.robot.libero.run_libero_eval \
      --config cosmos_predict2_2b_480p_libero__inference_only \
      --ckpt_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B \
      --config_file cosmos_policy/config/config.py \
      --use_wrist_image True \
      --use_proprio True \
      --normalize_proprio True \
      --unnormalize_actions True \
      --dataset_stats_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_dataset_statistics.json \
      --t5_text_embeddings_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_t5_embeddings.pkl \
      --trained_with_image_aug True \
      --chunk_size 16 \
      --num_open_loop_steps 16 \
      --task_suite_name "$suite" \
      --num_trials_per_task 50 \
      --local_log_dir cosmos_policy/experiments/robot/libero/logs/ \
      --seed 195 \
      --randomize_seed False \
      --deterministic True \
      --run_id_note "suite_${suite}" \
      --ar_future_prediction False \
      --ar_value_prediction False \
      --use_jpeg_compression True \
      --flip_images True \
      --num_denoising_steps_action 5 \
      --num_denoising_steps_future_state 1 \
      --num_denoising_steps_value 1 \
      --data_collection False
done

log_files = sorted(glob.glob("cosmos_policy/experiments/robot/libero/logs/**/*.json", recursive=True)) with open(log_files[-1]) as f: results = json.load(f)

print(results)


**步骤5：扩展评估规模**

覆盖所有四个LIBERO任务集，每个任务运行50次试跑：

```bash
for suite in libero_spatial libero_object libero_goal libero_10; do
  uv run --extra cu128 --group libero --python 3.10 \
    python -m cosmos_policy.experiments.robot.libero.run_libero_eval \
      --config cosmos_predict2_2b_480p_libero__inference_only \
      --ckpt_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B \
      --config_file cosmos_policy/config/config.py \
      --use_wrist_image True \
      --use_proprio True \
      --normalize_proprio True \
      --unnormalize_actions True \
      --dataset_stats_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_dataset_statistics.json \
      --t5_text_embeddings_path nvidia/Cosmos-Policy-LIBERO-Predict2-2B/libero_t5_embeddings.pkl \
      --trained_with_image_aug True \
      --chunk_size 16 \
      --num_open_loop_steps 16 \
      --task_suite_name "$suite" \
      --num_trials_per_task 50 \
      --local_log_dir cosmos_policy/experiments/robot/libero/logs/ \
      --seed 195 \
      --randomize_seed False \
      --deterministic True \
      --run_id_note "suite_${suite}" \
      --ar_future_prediction False \
      --ar_value_prediction False \
      --use_jpeg_compression True \
      --flip_images True \
      --num_denoising_steps_action 5 \
      --num_denoising_steps_future_state 1 \
      --num_denoising_steps_value 1 \
      --data_collection False
done

Workflow 2: RoboCasa evaluation

工作流2：RoboCasa评估

Copy this checklist and track progress:

text

RoboCasa Eval Progress:
- [ ] Step 1: Install RoboCasa assets and verify macros
- [ ] Step 2: Run single-task smoke evaluation
- [ ] Step 3: Validate outputs
- [ ] Step 4: Expand to multi-task runs

Step 1: Install RoboCasa

bash

git clone https://github.com/moojink/robocasa-cosmos-policy.git
uv pip install -e robocasa-cosmos-policy
python -m robocasa.scripts.setup_macros
python -m robocasa.scripts.download_kitchen_assets

This fork installs the

robocasa

Python package expected by Cosmos Policy while preserving the patched environment changes used in the public RoboCasa eval path. Verify

macros_private.py

exists and paths are correct.

Step 2: Single-task smoke evaluation

bash

uv run --extra cu128 --group robocasa --python 3.10 \
  python -m cosmos_policy.experiments.robot.robocasa.run_robocasa_eval \
    --config cosmos_predict2_2b_480p_robocasa_50_demos_per_task__inference \
    --ckpt_path nvidia/Cosmos-Policy-RoboCasa-Predict2-2B \
    --config_file cosmos_policy/config/config.py \
    --use_wrist_image True \
    --num_wrist_images 1 \
    --use_proprio True \
    --normalize_proprio True \
    --unnormalize_actions True \
    --dataset_stats_path nvidia/Cosmos-Policy-RoboCasa-Predict2-2B/robocasa_dataset_statistics.json \
    --t5_text_embeddings_path nvidia/Cosmos-Policy-RoboCasa-Predict2-2B/robocasa_t5_embeddings.pkl \
    --trained_with_image_aug True \
    --chunk_size 32 \
    --num_open_loop_steps 16 \
    --task_name TurnOffMicrowave \
    --obj_instance_split A \
    --num_trials_per_task 2 \
    --local_log_dir cosmos_policy/experiments/robot/robocasa/logs/ \
    --seed 195 \
    --randomize_seed False \
    --deterministic True \
    --run_id_note smoke \
    --use_variance_scale False \
    --use_jpeg_compression True \
    --flip_images True \
    --num_denoising_steps_action 5 \
    --num_denoising_steps_future_state 1 \
    --num_denoising_steps_value 1 \
    --data_collection False

Step 3: Validate outputs

Confirm the eval log prints the expected task name, object split, and checkpoint/config values.
Inspect the final
```
Success rate:
```
line in the log.

Step 4: Expand scope

Increase

--num_trials_per_task

or add more tasks. Keep

--obj_instance_split

fixed across repeated runs for comparability.

复制此检查清单跟踪进度：

text

RoboCasa评估进度:
- [ ] 步骤1：安装RoboCasa资产并验证宏配置
- [ ] 步骤2：运行单任务冒烟评估
- [ ] 步骤3：验证输出
- [ ] 步骤4：扩展为多任务运行

步骤1：安装RoboCasa

bash

git clone https://github.com/moojink/robocasa-cosmos-policy.git
uv pip install -e robocasa-cosmos-policy
python -m robocasa.scripts.setup_macros
python -m robocasa.scripts.download_kitchen_assets

此分支会安装Cosmos Policy所需的

robocasa

Python包，同时保留公开RoboCasa评估路径中使用的补丁环境变更。请验证

macros_private.py

存在且路径正确。

步骤2：单任务冒烟评估

bash

uv run --extra cu128 --group robocasa --python 3.10 \
  python -m cosmos_policy.experiments.robot.robocasa.run_robocasa_eval \
    --config cosmos_predict2_2b_480p_robocasa_50_demos_per_task__inference \
    --ckpt_path nvidia/Cosmos-Policy-RoboCasa-Predict2-2B \
    --config_file cosmos_policy/config/config.py \
    --use_wrist_image True \
    --num_wrist_images 1 \
    --use_proprio True \
    --normalize_proprio True \
    --unnormalize_actions True \
    --dataset_stats_path nvidia/Cosmos-Policy-RoboCasa-Predict2-2B/robocasa_dataset_statistics.json \
    --t5_text_embeddings_path nvidia/Cosmos-Policy-RoboCasa-Predict2-2B/robocasa_t5_embeddings.pkl \
    --trained_with_image_aug True \
    --chunk_size 32 \
    --num_open_loop_steps 16 \
    --task_name TurnOffMicrowave \
    --obj_instance_split A \
    --num_trials_per_task 2 \
    --local_log_dir cosmos_policy/experiments/robot/robocasa/logs/ \
    --seed 195 \
    --randomize_seed False \
    --deterministic True \
    --run_id_note smoke \
    --use_variance_scale False \
    --use_jpeg_compression True \
    --flip_images True \
    --num_denoising_steps_action 5 \
    --num_denoising_steps_future_state 1 \
    --num_denoising_steps_value 1 \
    --data_collection False

步骤3：验证输出

确认评估日志打印了预期的任务名、对象拆分和检查点/配置值
检查日志中最终的
```
Success rate:
```
行

步骤4：扩大评估范围

增加

--num_trials_per_task

参数或添加更多任务。重复运行时请固定

--obj_instance_split

参数以保证结果可比性。

Workflow 3: Blank-machine cluster launch

工作流3：裸机集群启动

text

Cluster Launch Progress:
- [ ] Step 1: Clone the public repo and enter the supported runtime
- [ ] Step 2: Sync the benchmark-specific dependency group
- [ ] Step 3: Export rendering and cache environment variables before eval

Step 1: Clone and enter the supported runtime

bash

git clone https://github.com/NVlabs/cosmos-policy.git
cd cosmos-policy

text

集群启动进度:
- [ ] 步骤1：克隆公开仓库并进入支持的运行时环境
- [ ] 步骤2：同步基准测试专属依赖组
- [ ] 步骤3：评估前导出渲染和缓存环境变量

步骤1：克隆仓库并进入支持的运行时

bash

git clone https://github.com/NVlabs/cosmos-policy.git
cd cosmos-policy

Follow SETUP.md, start the Docker container, and enter it before continuing.

参考SETUP.md启动Docker容器并进入后再继续后续操作


**Step 2: Sync dependencies**

```bash
uv sync --extra cu128 --group libero --python 3.10


**步骤2：同步依赖**

```bash
uv sync --extra cu128 --group libero --python 3.10

or, for RoboCasa:

如果是RoboCasa评估则执行：

uv sync --extra cu128 --group robocasa --python 3.10

then install the Cosmos-compatible RoboCasa fork:

然后安装兼容Cosmos的RoboCasa分支：

git clone https://github.com/moojink/robocasa-cosmos-policy.git uv pip install -e robocasa-cosmos-policy


**Step 3: Export runtime environment**

```bash
export CUDA_VISIBLE_DEVICES=0
export MUJOCO_EGL_DEVICE_ID=0
export MUJOCO_GL=egl
export PYOPENGL_PLATFORM=egl
export HF_HOME=${HF_HOME:-$HOME/.cache/huggingface}
export TRANSFORMERS_CACHE=${TRANSFORMERS_CACHE:-$HF_HOME}

git clone https://github.com/moojink/robocasa-cosmos-policy.git uv pip install -e robocasa-cosmos-policy


**步骤3：导出运行时环境变量**

```bash
export CUDA_VISIBLE_DEVICES=0
export MUJOCO_EGL_DEVICE_ID=0
export MUJOCO_GL=egl
export PYOPENGL_PLATFORM=egl
export HF_HOME=${HF_HOME:-$HOME/.cache/huggingface}
export TRANSFORMERS_CACHE=${TRANSFORMERS_CACHE:-$HF_HOME}

Expected performance benchmarks

预期性能基准

Reference values from official evaluation (tied to specific setup and seeds):

Task Suite	Success Rate	Notes
LIBERO-Spatial	98.1%	Official LIBERO spatial result
LIBERO-Object	100.0%	Official LIBERO object result
LIBERO-Goal	98.2%	Official LIBERO goal result
LIBERO-Long	97.6%	Official LIBERO long-horizon result
LIBERO-Average	98.5%	Official average across LIBERO suites
RoboCasa	67.1%	Official RoboCasa average result

Reproduction note: Published success rates still depend on checkpoint choice, task suite, seeds, and simulator setup. Record the exact command and environment alongside any reported number.

官方评估的参考值（与特定配置和种子绑定）：

任务集	成功率	说明
LIBERO-Spatial	98.1%	官方LIBERO空间任务结果
LIBERO-Object	100.0%	官方LIBERO物体任务结果
LIBERO-Goal	98.2%	官方LIBERO目标任务结果
LIBERO-Long	97.6%	官方LIBERO长程任务结果
LIBERO-平均	98.5%	官方LIBERO所有任务集平均结果
RoboCasa	67.1%	官方RoboCasa平均结果

复现说明：公开的成功率依赖于检查点选择、任务集、种子和仿真器配置。上报任何数值时请同时记录准确的命令和环境信息。

Non-negotiable rules

硬性规则

EGL alignment: Always set

CUDA_VISIBLE_DEVICES

MUJOCO_EGL_DEVICE_ID

MUJOCO_GL=egl

, and

PYOPENGL_PLATFORM=egl

together on headless GPU nodes.

Official runtime first: If host-Python installs hit binary compatibility issues, fall back to the supported container workflow from
```
SETUP.md
```
before debugging package internals.
Cache consistency: Use the same cache directory across setup and eval so Hugging Face and dependency caches are reused.
Run comparability: Keep task name, object split, seed, and trial count fixed across repeated runs.

EGL对齐：在无桌面GPU节点上必须同时设置

CUDA_VISIBLE_DEVICES

、

MUJOCO_EGL_DEVICE_ID

、

MUJOCO_GL=egl

和

PYOPENGL_PLATFORM=egl

优先官方运行时：如果主机Python安装遇到二进制兼容性问题，请先回退到
```
SETUP.md
```
中支持的容器工作流，再调试包内部问题
缓存一致性：搭建环境和评估时使用相同的缓存目录，以复用Hugging Face和依赖缓存
运行可比性：重复运行时请固定任务名、对象拆分、种子和试跑次数

Common issues

常见问题

Issue: binary compatibility or loader failures on host Python

Fix: rerun inside the official container/runtime from

SETUP.md

. Do not assume host-package rebuilds will match the public release environment.

Issue: LIBERO prompts for config path in a non-interactive shell

Fix: pre-create

LIBERO_CONFIG_PATH/config.yaml

python

import os, yaml

config_dir = os.path.expanduser("~/.libero")
os.makedirs(config_dir, exist_ok=True)
with open(os.path.join(config_dir, "config.yaml"), "w") as f:
    yaml.dump({"benchmark_root": "/path/to/libero/datasets"}, f)

Issue: EGL initialization or shutdown noise

Fix: align EGL environment variables first. Treat teardown-only

EGL_NOT_INITIALIZED

warnings as low-signal unless the job exits non-zero.

Issue: Kitchen object sampling NaNs or asset lookup failures in RoboCasa

Fix: rerun asset setup and confirm the patched robocasa install is intact:

bash

python -m robocasa.scripts.download_kitchen_assets
python -c "import robocasa; print(robocasa.__file__)"

Issue: MuJoCo rendering mismatch

Fix: verify GPU device alignment:

python

import os
cuda_dev = os.environ.get("CUDA_VISIBLE_DEVICES", "not set")
egl_dev = os.environ.get("MUJOCO_EGL_DEVICE_ID", "not set")
assert cuda_dev == egl_dev, f"GPU mismatch: CUDA={cuda_dev}, EGL={egl_dev}"
print(f"Rendering on GPU {cuda_dev}")

问题：主机Python上出现二进制兼容性或加载器失败

解决方法：在

SETUP.md

提供的官方容器/运行时中重新运行。不要假设主机包重建会匹配公开发布的环境。

问题：LIBERO在非交互shell中提示输入配置路径

解决方法：预先创建

LIBERO_CONFIG_PATH/config.yaml

：

python

import os, yaml

config_dir = os.path.expanduser("~/.libero")
os.makedirs(config_dir, exist_ok=True)
with open(os.path.join(config_dir, "config.yaml"), "w") as f:
    yaml.dump({"benchmark_root": "/path/to/libero/datasets"}, f)

问题：EGL初始化或关闭报错

解决方法：先对齐EGL环境变量。如果任务没有非零退出，仅关闭阶段的

EGL_NOT_INITIALIZED

警告可以忽略。

问题：RoboCasa中厨房对象采样出现NaN或资产查找失败

解决方法：重新运行资产设置，确认打过补丁的robocasa安装完整：

bash

python -m robocasa.scripts.download_kitchen_assets
python -c "import robocasa; print(robocasa.__file__)"

问题：MuJoCo渲染不匹配

解决方法：验证GPU设备对齐：

python

import os
cuda_dev = os.environ.get("CUDA_VISIBLE_DEVICES", "not set")
egl_dev = os.environ.get("MUJOCO_EGL_DEVICE_ID", "not set")
assert cuda_dev == egl_dev, f"GPU mismatch: CUDA={cuda_dev}, EGL={egl_dev}"
print(f"Rendering on GPU {cuda_dev}")

Advanced topics

进阶主题

LIBERO command matrix: See references/libero-commands.md RoboCasa command matrix: See references/robocasa-commands.md

LIBERO命令矩阵：参考references/libero-commands.md RoboCasa命令矩阵：参考references/robocasa-commands.md

Resources

资源

Cosmos Policy repository: https://github.com/NVlabs/cosmos-policy
LIBERO benchmark: https://github.com/Lifelong-Robot-Learning/LIBERO
Cosmos-compatible RoboCasa fork: https://github.com/moojink/robocasa-cosmos-policy
Upstream RoboCasa project: https://github.com/robocasa/robocasa
MuJoCo documentation: https://mujoco.readthedocs.io/

Cosmos Policy仓库: https://github.com/NVlabs/cosmos-policy
LIBERO基准: https://github.com/Lifelong-Robot-Learning/LIBERO
兼容Cosmos的RoboCasa分支: https://github.com/moojink/robocasa-cosmos-policy
上游RoboCasa项目: https://github.com/robocasa/robocasa
MuJoCo文档: https://mujoco.readthedocs.io/