SatNav Baseline Model Zoo
Collection
Baseline checkpoints for SatNav vision-language navigation. This collection includes StreamVLN, OpenFly, UniNaVid, and NaVILA • 8 items • Updated
Uni-NaVid SatNav Scratch 1ep lr1e-5
This repository contains a Hugging Face upload-ready copy of the Uni-NaVid
SatNav scratch baseline uninavid-satnav-scratch-1ep-lr1e-5.
Model Details
- Model type: Uni-NaVid-style vision-language-navigation policy.
- Training mode: SatNav fine-tuning from
lmsys/vicuna-7b-v1.5. - Language backbone family: Vicuna/Llama 2.
- Vision tower source: EVA-CLIP checkpoint used by Uni-NaVid.
- Local archive source:
output/model_zoo/baseline/uninavid-baseline-scratch-1ep-data260418-bs192-lr1e-5-20260418-203618.
Repository Contents
This release directory keeps only Uni-NaVid inference/evaluation artifacts:
- tokenizer files
config.json- sharded PyTorch model weights
pytorch_model.bin.index.json- model card, license, and notice files
Training state, optimizer state, scheduler state, RNG state, trainer state, and
duplicated checkpoint-* directories are intentionally omitted.
Training Data and Procedure
The model was fine-tuned for 1 epoch on SatNav trajectory data using the
Uni-NaVid SatNav training pipeline. The default optimizer learning rate was
1e-5. The model card does not redistribute SatNav episodes, images, simulator
scenes, or training logs.
License
This release is marked with the Hugging Face llama2 license tag because the
released weights are fine-tuned from Vicuna/Llama 2-derived weights.
Uni-NaVid code, EVA-CLIP, SatNav assets, and other third-party components remain
under their respective upstream licenses and terms.
- Downloads last month
- 12
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Eku127/uninavid-satnav-scratch-1ep-lr1e-5
Base model
lmsys/vicuna-7b-v1.5