Eku127/swiftvln-satnav-qwen3vl-2b-1ep-f32s4-overlap0-pf-h8-pool-s2-noembed

SwiftVLN for SatNav — Qwen3-VL 2B

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 applies the SwiftVLN reference memory design to a Qwen3-VL 2B backbone.

Model

• —Starting checkpoint: Qwen/Qwen3-VL-2B-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
• —qwen3vl-2b: Qwen3-VL 2B backbone
• —1ep: trained for one epoch
• —f32s4: uses a 32-frame window and predicts four actions
• —overlap0: uses non-overlapping training windows
• —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, base-backbone comparison, row Qwen3-VL-2B.

SR, SPL, and OS are percentages. NE is measured in meters. Steps is the average number of executed environment actions, including turns and stop.

Usage

Download the checkpoint to a local directory and pass that directory through MODEL_PATH:

MODEL_NAME=swiftvln-satnav-qwen3vl-2b-1ep-f32s4-overlap0-pf-h8-pool-s2-noembed
MODEL_PATH=/path/to/${MODEL_NAME} \
bash scripts/eval/eval_by_name.sh "${MODEL_NAME}"

See the SwiftVLN evaluation guide for environment, data, and single- or multi-GPU setup. Keep the full repository name unchanged because SwiftVLN derives the evaluation configuration from it.