SwiftVLN for SatNav — Overlap 4

Related Repositories

• SwiftVLN: training and evaluation code for these checkpoints.
• SatNav: satellite-image navigation environments, datasets, and evaluation tools.

This checkpoint is designed for SwiftVLN on SatNav, the continuous-state vision-and-language navigation benchmark over satellite imagery. It studies a sliding-window setup in which adjacent training windows overlap by four steps.

Model

• Starting checkpoint: Qwen/Qwen2.5-VL-3B-Instruct
• Training data: SatNav-v0.1 offline expert trajectories
• Training: 1 epoch, full-parameter fine-tuning, learning rate 2e-5
• Context: 32 RGB frames, up to 8 uniformly sampled history frames
• Prediction horizon: 4 actions
• Memory: per-frame average pooling with stride 2

Repository name

• swiftvln-satnav: SwiftVLN trained and evaluated on SatNav
• 3b: Qwen2.5-VL 3B backbone
• 1ep: trained for one epoch
• f32s4: uses a 32-frame window and predicts four actions
• overlap4: adjacent training windows overlap by four steps
• pf-h8: uses per-frame memory with up to eight history frames
• pool-s2: applies average pooling with stride 2
• noembed: uses no embedding enhancement

Reference Results

Results reported in the SatNav paper, memory-design ablation, row Overlap turns (N_o = 2).

SR, SPL, and OS are percentages. Changes in SR and OS are percentage points relative to the SwiftVLN reference model.

Usage

Use this checkpoint with the SwiftVLN evaluation guide. Keep the full repository name unchanged because SwiftVLN derives the evaluation configuration from it.

This checkpoint is a sliding-window ablation in the SwiftVLN SatNav Ablation Model Zoo.