Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+mmproj-F16.gguf filter=lfs diff=lfs merge=lfs -text
+Qwen3.5-2B_F16.gguf filter=lfs diff=lfs merge=lfs -text
+Qwen3.5-2B_Q4_k_m.gguf filter=lfs diff=lfs merge=lfs -text
+Qwen3.5-2B_Q5_k_m.gguf filter=lfs diff=lfs merge=lfs -text

Qwen3.5-2B_F16.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9494e489554a5efe5d0ec28113753ac82ef19da7901447ced7d6fc59c5591a35
+size 3775709056

Qwen3.5-2B_Q4_k_m.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:516a5ce4d26131c7e8278b57fa3ee877ca85c98a8a2bbe8ff0fd52aa8cc4322b
+size 1270808448

Qwen3.5-2B_Q5_k_m.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1b64287fb968709493e963aa025ff04d0775ba82061c5f5e53c48ba5914a7e16
+size 1424768896

README.md

@@ -0,0 +1,143 @@
+---
+license: apache-2.0
+language:
+- en
+- zh
+base_model: Qwen/Qwen3.5-2B
+tags:
+  - image-text-to-text
+  - vision-language
+  - multimodal
+  - reasoning
+  - long-context
+  - multilingual
+  - lightweight
+---
+
+## Qwen3.5-2B
+
+Qwen3.5-2B is a compact vision-language model from the Qwen 3.5 series developed by Alibaba Cloud. The model is designed to handle multimodal inputs where images and text prompts can be combined to generate informative textual responses.
+
+With approximately 2 billion parameters, the model balances performance and efficiency, enabling multimodal reasoning and visual understanding while remaining suitable for deployment on modest hardware. The model can analyze images, diagrams, screenshots, and documents and produce contextual explanations or answers based on the provided prompt.
+
+The Qwen3.5 small series focuses on efficient models optimized for research, experimentation, and practical deployment scenarios where large models may be unnecessary or computationally expensive.
+
+---
+
+## Model Overview
+
+- **Model Name**: Qwen3.5-2B
+- **Base Model**: Qwen3.5-2B
+- **Architecture**: Multimodal Transformer (Vision Encoder + Language Model)
+- **Parameter Count**: ~2 Billion
+- **Context Window**: Up to ~256K tokens (implementation dependent)
+- **Modalities**: Image, Text
+- **Primary Languages**: English, Chinese, multilingual capability
+- **Developer**: Qwen (Alibaba Cloud)
+- **License**: Apache 2.0
+
+---
+
+## Quantization Details
+
+### FP16
+
+- Approx. ~65% size reduction compared to FP16
+- Very low memory footprint (~ 1.18 GB)
+- Highest fidelity to pretrained weights
+- Recommended for GPU inference and evaluation workloads
+
+### Q4_K_M
+
+- Approx. ~60% size reduction with higher fidelity (~ 1.33 GB)
+- Slightly larger size than Q4_K_M 
+- Designed for efficient inference on consumer hardware
+- Compatible with CPU inference and low-VRAM GPUs
+
+---
+
+## Training Overview
+
+### Pretraining
+
+The model is pretrained on large-scale multimodal datasets containing paired image–text data together with extensive textual corpora. This training enables the model to learn strong associations between visual features and natural language representations.
+
+Training objectives include:
+
+- Visual–text alignment
+- Multimodal representation learning
+- Language modeling and reasoning
+- Cross-modal understanding
+
+### Optimization
+
+Additional optimization stages improve the model’s ability to perform multimodal tasks such as:
+
+- Visual question answering
+- Image caption generation
+- Scene and object recognition
+- Chart and document interpretation
+
+---
+
+## Core Capabilities
+
+- **Multimodal understanding**
+  Processes both image and text inputs to produce meaningful responses.
+
+- **Visual question answering**
+  Interprets visual content and answers questions about objects, scenes, or diagrams.
+
+- **Image captioning**
+  Generates descriptive captions explaining the contents of images.
+
+- **Image-grounded reasoning**
+  Performs reasoning tasks using information extracted from visual inputs.
+
+- **Multilingual interaction**
+  Supports multiple languages, with strong English and Chinese performance.
+
+- **Long-context processing**
+  Capable of handling extended inputs and longer multimodal conversations.
+
+---
+
+## Example Usage
+
+### llama.cpp
+
+```
+./llama-cli \
+  -m SandlogicTechnologies\Qwen3.5-2B_Q4_K_M.gguf \
+  -p "What is Knowledge Distillation?"
+```
+
+---
+
+## Recommended Use Cases
+
+- Multimodal conversational assistants
+- Visual question answering systems
+- Document and screenshot analysis
+- Chart and diagram interpretation
+- Image captioning and visual description
+- Educational tools using visual materials
+- Research involving multimodal reasoning
+- Rapid prototyping of multimodal AI applications
+
+---
+
+## Acknowledgments
+
+These quantized models are based on the original work by **Qwen** development team.
+
+Special thanks to:
+
+- The [Qwen](https://huggingface.co/Qwen) team for developing and releasing the [Qwen3.5-2B](https://huggingface.co/Qwen/Qwen3.5-2B) model.
+- **Georgi Gerganov** and the `llama.cpp` community for enabling efficient inference using the GGUF format.
+
+---
+
+## Contact
+
+For any inquiries or support, please contact us at support@sandlogic.com or visit our [Website](https://www.sandlogic.com/).

mmproj-F16.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7035e9cb8d7c6a9681d07eef9a364783e86ea4cd73faab2eabb4f43a101830c7
+size 668227264