Bump to v0.3.0

[kew6688]

Motivation

Please describe the motivation of this PR and the goal you want to achieve through this PR.

Modification

Bump to v0.3.0

BC-breaking (Optional)

Major breaks in InternVLA-N1 usage:

• in internvla-n1 policy InternVLAN1Net: policy.s1_step_latent() discard use_async parameter
• in internvla-n1 model InternVLAN1ForCausalLM: model.generate_traj() discard use_async parameter

Use cases (Optional)

• Add InternVLA-N1 Training code
• Support RxR dataset evaluation
• Add the VL-LN benchmark, including dataset, trainer, evaluator, dialog agent
• Add a new controller - Flash controller with collision check

Changelog of v0.3.0 (2025/12/30)

Highlights

• Add InternVLA-N1 Training code ([Feature] Add training code for InternVLA-N1 #184)
• Support training and evaluation for the VL-LN benchmark ([Feature] Support Evaluation in VL-LN Bench #193, [Feature] Add training code for VLLN #198)
• Add a new flash without collision controller ([Feature] Support Flash without collision Controller #189)

New Features

• Support InternVLA-N1 Training code ([Feature] Add training code for InternVLA-N1 #184)
• Support RxR dataset evaluation ([Feature] Add training code for InternVLA-N1 #184)
• Add training code for the baseline of VLLN Bench ([Feature] Add training code for VLLN #198)
• Support evaluation in VL-LN Bench ([Feature] Support Evaluation in VL-LN Bench #193)
• Support Flash without collisoin controller ([Feature] Support Flash without collision Controller #189)

Improvements

• Decouple System 2 and Dual System evaluation functions in Habitat evaluator for better readability ([Feature] Add training code for InternVLA-N1 #184)
• Update internvla-n1 agent in VLN-PE to align with new model and policy interface ([Feature] Add training code for InternVLA-N1 #184)
• Enhance Habitat evaluation pipeline to handle NaN values in results ([Fix] Fix issues noticed during Regression benchmark #217)
• Update readme to include community tutorials ([Fix] Fix issues noticed during Regression benchmark #217)

Bug Fixes

• Fix the version of diffusers in requirements ([Feature] Add training code for InternVLA-N1 #184)
• Fix the result json saving path in VLN-PE ([Fix] Fix issues noticed during Regression benchmark #217)
• Fix RxR evaluation result collecting bug ([Fix] Fix issues noticed during Regression benchmark #217)
• Removed legacy code in scripts/demo ([Fix] Fix issues noticed during Regression benchmark #217)

Contributors

@kellyiss @DuangZhu @0309hws @kew6688

[Tai-Wang]

Need to add the reference for doc of using InternVLA-N1 training code