reaperdoesntknow/Qwen3.5-2B-CyberSec

VISIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:2.3BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 9, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

reaperdoesntknow/Qwen3.5-2B-CyberSec is a 2.3 billion parameter Qwen3.5 model fine-tuned by reaperdoesntknow for cybersecurity instruction prompts, leveraging the Trendyol Cybersecurity Instruction Tuning Dataset. This model, with a 32768 token context length, is designed for research and local experimentation in cybersecurity-related natural language processing tasks. It features a conditional-generation architecture with text and vision components, making it suitable for exploring small-model responses to specialized cybersecurity queries.

Loading preview...

Overview

reaperdoesntknow/Qwen3.5-2B-CyberSec is a 2.3 billion parameter Qwen3.5 model, fine-tuned using the Trendyol Cybersecurity Instruction Tuning Dataset. This model is specifically adapted for processing and responding to cybersecurity-related prompts, building upon the unsloth/Qwen3.5-2B base model. It supports a 32768 token context length and is provided in Transformers / Safetensors format, featuring both text and vision components.

Key Capabilities

  • Cybersecurity Instruction Tuning: Specialized for understanding and generating responses to cybersecurity-specific instructions.
  • Research and Experimentation: Intended for academic research, local prototyping, and qualitative evaluation of small models in cybersecurity contexts.
  • Multimodal Architecture: Configured with a conditional-generation architecture that includes both text and vision capabilities.
  • Base Model Comparison: Useful for comparing performance against the upstream Qwen3.5 2B checkpoint.

Intended Use Cases

  • Research: Investigating how small models respond to cybersecurity instruction prompts.
  • Prototyping: Developing and testing cybersecurity-focused applications locally.
  • Evaluation: Qualitative assessment and comparison with other models in the cybersecurity domain.
  • Conversion Experiments: Suitable for experiments involving model conversion and quantization (e.g., GGUF builds).

Limitations and Safety

It is crucial to note that this model has not undergone formal benchmark or safety evaluations. It may generate incorrect or unsafe technical guidance, and its training data could contain errors or outdated practices. Users should exercise caution, review all generated outputs, and not rely on the model as the sole basis for critical security decisions.

Finetunes
2 models