---
title: OmniNav_Flow
canonical_url: "https://www.modelscope.ai/models/chongchongjj/OmniNav_Flow"
md_url: "https://www.modelscope.ai/models/chongchongjj/OmniNav_Flow.md"
repository: chongchongjj/OmniNav_Flow
last_updated: 2026-09-11
license: apache-2.0
pipeline_tag: image-text-to-text
tasks:
  - image-text-to-text
model_type:
  - qwen2_5_vl
architectures:
  - Qwen2_5_VLForConditionalGeneration
parameters: 4.7B
tensor_type:
  - BF16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 55
stars: 0
---

# OmniNav_Flow

> OmniNav_Flow - An open-source model by chongchongjj on ModelScope. Flow Matching inference checkpoint for OmniNav on R2R, RxR, and OVON.

chongchongjj/OmniNav_Flow is a 4.7B-parameter image-text-to-text model on ModelScope. licensed under apache-2.0.

- **Repository**: chongchongjj/OmniNav_Flow
- **License**: apache-2.0
- **Tasks**: image-text-to-text
- **Parameters**: 4.7B
- **Downloads**: 55
- **Stars**: 0
- **Last updated**: 2026-09-11

Source: https://www.modelscope.ai/models/chongchongjj/OmniNav_Flow

---

# OmniNav_Flow

Flow Matching inference checkpoint for [OmniNav](https://github.com/amap-cvlab/OmniNav), supporting R2R, RxR, and OVON navigation evaluation.

## Inference

Use this checkpoint with the Flow Matching inference branch in OmniNav:

```bash
FLOW_MATCH=true bash eval_r2r.sh
FLOW_MATCH=true bash eval_rxr.sh
FLOW_MATCH=true bash eval_ovon.sh
```

Set `MODEL_PATH` in the corresponding evaluation script to the downloaded checkpoint directory.

This repository contains inference model weights and the tokenizer/processor configuration required to load them. Optimizer states, gradients, trainer states, random-number-generator states, training arguments, and dataset configurations are not included.