Update README.md

README.md

@@ -1,3 +1,79 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+---
+
+# **Spatial-VU**
+
+> The **Spatial-VU** model is a fine-tuned variant of **Qwen2.5-VL-7B-Instruct**, developed for **Spatial Reasoning** and **Vision Understanding**. It is designed to deliver detailed, context-aware visual descriptions and reasoning outputs across a wide range of image types, resolutions, and aspect ratios.
+
+## Key Highlights
+
+* **Spatial Reasoning and Visual Comprehension**: Optimized for interpreting spatial layouts, object relationships, and scene understanding within images.
+* **High-Precision Descriptions**: Generates detailed, context-rich captions for general, technical, and abstract imagery.
+* **Adaptive Across Aspect Ratios**: Performs effectively with images of varying formats—wide, tall, square, and irregular.
+* **Multi-Level Detail Control**: Supports both concise summaries and fine-grained analytical outputs, depending on the prompt.
+* **Built on Qwen2.5-VL Architecture**: Utilizes the visual-linguistic reasoning power of Qwen2.5-VL-7B for structured and accurate comprehension tasks.
+* **Multilingual Support**: Outputs in English by default, with the ability to generate multilingual responses through prompt conditioning.
+
+## Quick Start with Transformers
+
+```python
+from transformers import Qwen2_5_VLForConditionalGeneration, AutoTokenizer, AutoProcessor
+from qwen_vl_utils import process_vision_info
+
+model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
+    "prithivMLmods/Spatial-VU", torch_dtype="auto", device_map="auto"
+)
+
+processor = AutoProcessor.from_pretrained("prithivMLmods/Spatial-VU")
+
+messages = [
+    {
+        "role": "user",
+        "content": [
+            {
+                "type": "image",
+                "image": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg",
+            },
+            {"type": "text", "text": "Analyze the spatial layout and describe the scene."},
+        ],
+    }
+]
+
+text = processor.apply_chat_template(
+    messages, tokenize=False, add_generation_prompt=True
+)
+image_inputs, video_inputs = process_vision_info(messages)
+inputs = processor(
+    text=[text],
+    images=image_inputs,
+    videos=video_inputs,
+    padding=True,
+    return_tensors="pt",
+)
+inputs = inputs.to("cuda")
+
+generated_ids = model.generate(**inputs, max_new_tokens=128)
+generated_ids_trimmed = [
+    out_ids[len(in_ids):] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
+]
+output_text = processor.batch_decode(
+    generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
+)
+print(output_text)
+```
+
+## Intended Use
+
+* Spatial reasoning, visual understanding, and scene analysis tasks.
+* Descriptive and interpretive caption generation for research or vision-language evaluation.
+* Structured visual comprehension in creative and analytical applications.
+* Data annotation and reasoning augmentation in multimodal AI pipelines.
+* Spatial layout and context interpretation across diverse visual domains.
+
+## Limitations
+
+* May generate variable outputs depending on the phrasing of prompts.
+* Not recommended for production use cases requiring strong moderation or controlled tone.
+* Performance may vary on abstract, heavily stylized, or low-context visual inputs.
+* Lacks fine-tuned control over subjective or ambiguous visual content interpretations.