🚀 Hume-Libero_Object模型卡片
Hume-Libero_Object是一個在Libero-Object數據集上訓練的雙系統視覺-語言-動作模型,具備系統2思維能力。它能在機器人領域發揮重要作用,為相關研究和應用提供有力支持。
🚀 快速開始
如果你想復現論文中的結果,請遵循此說明。
如果你想直接使用該模型,請參考以下代碼示例:
from hume import HumePolicy
import numpy as np
hume = HumePolicy.from_pretrained("/path/to/checkpoints")
hume.init_infer(
infer_cfg=dict(
replan_steps=8,
s2_replan_steps=16,
s2_candidates_num=5,
noise_temp_lower_bound=1.0,
noise_temp_upper_bound=1.0,
time_temp_lower_bound=0.9,
time_temp_upper_bound=1.0,
post_process_action=True,
device="cuda",
)
)
observation = {
"observation.images.image": np.zeros((1,224,224,3), dtype = np.uint8),
"observation.images.wrist_image": np.zeros((1,224,224,3), dtype = np.uint8),
"observation.state": np.zeros((1, 7)),
"task": ["Lift the papper"],
}
action = hume.infer(observation)
💻 使用示例
基礎用法
from hume import HumePolicy
import numpy as np
hume = HumePolicy.from_pretrained("/path/to/checkpoints")
hume.init_infer(
infer_cfg=dict(
replan_steps=8,
s2_replan_steps=16,
s2_candidates_num=5,
noise_temp_lower_bound=1.0,
noise_temp_upper_bound=1.0,
time_temp_lower_bound=0.9,
time_temp_upper_bound=1.0,
post_process_action=True,
device="cuda",
)
)
observation = {
"observation.images.image": np.zeros((1,224,224,3), dtype = np.uint8),
"observation.images.wrist_image": np.zeros((1,224,224,3), dtype = np.uint8),
"observation.state": np.zeros((1, 7)),
"task": ["Lift the papper"],
}
action = hume.infer(observation)
高級用法
from hume import HumePolicy
import numpy as np
hume = HumePolicy.from_pretrained("/path/to/checkpoints")
hume.init_infer(
infer_cfg=dict(
replan_steps=8,
s2_replan_steps=16,
s2_candidates_num=5,
noise_temp_lower_bound=1.0,
noise_temp_upper_bound=1.0,
time_temp_lower_bound=0.9,
time_temp_upper_bound=1.0,
post_process_action=True,
device="cuda",
)
)
observation = {
"observation.images.image": np.zeros((1,224,224,3), dtype = np.uint8),
"observation.images.wrist_image": np.zeros((1,224,224,3), dtype = np.uint8),
"observation.state": np.zeros((1, 7)),
"task": ["Lift the papper"],
}
action = hume.infer(observation)
📄 許可證
本項目採用MIT許可證。
📚 引用
如果你使用了該模型,請引用以下論文:
@article{song2025hume,
title={Hume: Introducing System-2 Thinking in Visual-Language-Action Model},
author={Anonimous Authors},
journal={arXiv preprint arXiv:2505.21432},
year={2025}
}
📋 模型信息