Issues

• [Reproduction] LIBERO-Long success rate ~77.8% when training from scratch with the provided recipe (vs ~97% from released weights)

  Robbyant/lingbot-va#121 · Filter by author VEGA1215 opened on Sep 21, 2026

• Post-training: torch 2.9.0 vs lerobot 0.3.3 conflict

  Robbyant/lingbot-va#117 · Filter by author 883-hurry opened on Aug 24, 2026

• Low success rate (19.4%) when evaluating official LIBERO checkpoint on LIBERO-Plus

  Robbyant/lingbot-va#115 · Filter by author Maxwell-Zhao opened on Aug 19, 2026

• Timeline for LingBot-VA 2.0 code release

  Robbyant/lingbot-va#114 · Filter by author Riecacc opened on Aug 7, 2026

• Slight gap when reproducing LIBERO-Long with the released checkpoint (97.2% vs. 98.5%)

  Robbyant/lingbot-va#112 · Filter by author daizzzR opened on Jul 23, 2026

• LIBERO Post-Training+Evaluation (all in one OR each)

  Robbyant/lingbot-va#111 · Filter by author Lourisy opened on Jul 21, 2026

• Request for Official LingBot-VA Evaluation on RMBench

  Robbyant/lingbot-va#110 · Filter by author Gucci233 opened on Jul 21, 2026
  2

• Clarification on latent actions ℓ_t vs. physical action chunks a_t in TI2VA/ICL training

  Robbyant/lingbot-va#109 · Filter by author DocZbs opened on Jul 21, 2026

• Verify evals on Papers with Code

  Robbyant/lingbot-va#106 · Filter by author NielsRogge opened on Jul 17, 2026

• Inquery about the version MoT and version 2.0

  Robbyant/lingbot-va#105 · Filter by author Victor-RoboNeo opened on Jul 13, 2026

• Inconsistent evaluation success rates on LIBERO with identical setup

  Robbyant/lingbot-va#104 · Filter by author changdong1687 opened on Jul 8, 2026

• reproducing pi0.5 numbers

  Robbyant/lingbot-va#103 · Filter by author quincy-u opened on Jul 6, 2026

• Cleaned & augmented RoboTwin dataset in LeRobot format for post-training

  Robbyant/lingbot-va#102 · Filter by author XiaoJichun opened on Jul 4, 2026

• Could you share the pretraining config？

  Robbyant/lingbot-va#101 · Filter by author chuanye-Wang opened on Jun 30, 2026

• KV Cache update strategy on environments with kinematic constraints

  Robbyant/lingbot-va#100 · Filter by author chuanye-Wang opened on Jun 30, 2026

• offloading VAE to CPU makes inference latency unacceptable

  Robbyant/lingbot-va#99 · Filter by author CCSCovenant opened on Jun 24, 2026

• DreamX-World (https://github.com/AMAP-ML/DreamX-World)

  Robbyant/lingbot-va#98 · Filter by author jiashu-zhu opened on Jun 18, 2026

• RoboTwin inference fails with frame_chunk_size=2 in Wan VAE temporal convolution

  Robbyant/lingbot-va#97 · Filter by author ww249 opened on Jun 18, 2026

• Training data for RoboTwin 2.0 — clean-only (2,500) vs. full (27,500)?

  Robbyant/lingbot-va#96 · Filter by author brahmatih opened on Jun 17, 2026
  3

• Real-World Exp hyperparameters

  Robbyant/lingbot-va#95 · Filter by author armanakbari opened on Jun 15, 2026
  1

• MoT Version

  Robbyant/lingbot-va#94 · Filter by author arashakb opened on Jun 12, 2026

• Infer chunk cold start inference time

  Robbyant/lingbot-va#93 · Filter by author PF-CR opened on Jun 12, 2026

• Discrepancies between codebase and paper (Denoising steps & MoT architecture)

  Robbyant/lingbot-va#92 · Filter by author mcddddz opened on Jun 1, 2026

• Regarding the version of PyTorch

  Robbyant/lingbot-va#89 · Filter by author KeliangLiu-Lab opened on May 27, 2026

• Error: LoweringException: AssertionError: target: flex_attention

  Robbyant/lingbot-va#87 · Filter by author COST-97 opened on May 22, 2026
  3