---
library_name: transformers
pipeline_tag: image-text-to-text
tags:
  - robotics
  - vision-language-action
  - manipulation
  - qwen3_5
---

# guava-v13b-qwen3.5-4b

A Qwen3.5-VL 4B model fine-tuned on the **Guava SFT data** robot-manipulation dataset
(2,268 episodes across 16 tabletop tasks). 

Training data: [AIcell/Guava-4B-SFT-data](https://huggingface.co/datasets/AIcell/Guava-4B-SFT-data).

## Model

| | |
|---|---|
| Architecture | `Qwen3_5ForConditionalGeneration` (vision + text) |
| Precision | bfloat16 |
| Hidden size | 2560, 32 layers (linear attention, full attention every 4th) |
| Vocab | 248,320 |
| Max positions | 262,144 |
| Vision | 24-layer ViT, patch 16, merge 2 |
| Shards | 2 safetensors, 8.47 GiB total |

## Usage

```python
from transformers import AutoModelForImageTextToText, AutoProcessor

model = AutoModelForImageTextToText.from_pretrained(
    "AIcell/Guava-4B", dtype="bfloat16", device_map="auto"
)
processor = AutoProcessor.from_pretrained("AIcell/Guava-4B")
```

Episodes in the training data contain up to 30 images, and coordinates are already
table-aligned (tabletop is z = 0) — do not apply table-height normalization a second time.

## Tasks

`apple_juice_order`, `bread_near_lemon`, `can_in_bin`, `close_drawer`, `cube_stack`,
`cube_under_cup`, `hotdog_near_donut`, `milk_near_cup`, `open_drawer`, `pick_up_orange`,
`push_basket`, `push_cereal`, `red_objects_in_basket`, `remove_cube_from_tray`,
`shell_game`, `tomato_in_bowl`