lookarooka's picture
Upload Humanoid-v5 PPO weights (Ep 183, 3460 steps)
f96e1fb verified
|
Raw History Blame Contribute Delete
1.12 kB
metadata
language:
  - en
  - ko
license: mit
tags:
  - reinforcement-learning
  - physical-ai
  - robotics
  - gymnasium
  - mujoco
  - humanoid-v5
  - ppo
  - actor-critic
pipeline_tag: robotics

πŸ€– Physical AI: Humanoid-v5 PPO Policy Model

λ³Έ λͺ¨λΈμ€ 피지컬 AI & Gymnasium κ°•ν™”ν•™μŠ΅(RL) μŠ€νŠœλ””μ˜€μ—μ„œ ν•™μŠ΅λœ MuJoCo Humanoid-v5 3D 보행 λ‘œλ΄‡ 신경망 μ²΄ν¬ν¬μΈνŠΈμž…λ‹ˆλ‹€.

πŸ“Š λͺ¨λΈ 메타데이터

  • ν™˜κ²½ (Environment): MuJoCo Humanoid-v5 (3D Physics)
  • ν•™μŠ΅ μ—ν”Όμ†Œλ“œ: 183
  • 총 νƒ€μž„μŠ€ν…: 3460
  • μ•Œκ³ λ¦¬μ¦˜: Continuous Actor-Critic (PPO)
  • 앑좔에이터/κ΄€μ ˆ 수: 17 continuous joints
  • κ΄€μΈ‘ 차원: 348-dim (45-dim core state)
  • μ—…λ‘œλ“œ μΌμ‹œ: 2026. 8. 28. μ˜€μ „ 10:28:40

πŸ“ μ €μž₯ 파일

  • model_weights.json: 17-앑좔에이터 Actor & Critic 신경망 κ°€μ€‘μΉ˜ 및 편ν–₯κ°’

πŸš€ λͺ¨λΈ λ‘œλ“œ 방법 (Physical AI Studio λŒ€μ‹œλ³΄λ“œ)

  1. Physical AI Studio μ›Ή λŒ€μ‹œλ³΄λ“œ μ‹€ν–‰
  2. [λͺ¨λΈ κ°€μ€‘μΉ˜ μ €μž₯ 및 파일 관리] μΉ΄λ“œμ—μ„œ [κ°€μ€‘μΉ˜ 파일 뢈러였기 (.json)] 클릭 ν›„ model_weights.json 선택