OMCHOKSI108/Paralay1.1-Merged

TEXT GENERATIONConcurrency Cost:1Model Size:1.5BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Jun 14, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

OMCHOKSI108/Paralay1.1-Merged is a 1.5 billion parameter instruction-tuned causal language model developed by Om Choksi, built upon Qwen2.5-1.5B-Instruct. This model is specifically fine-tuned using LoRA for defensive cybersecurity assistance, excelling in areas like incident response, threat modeling, and secure coding. It features a 32768 token context length and is designed to provide safe, factual guidance for security analysts, students, and developers.

Loading preview...

Paralay 1.1 Merged: A Specialized Defensive Cybersecurity Assistant

OMCHOKSI108/Paralay1.1-Merged is a 1.5 billion parameter large language model, fine-tuned by Om Choksi from the Qwen2.5-1.5B-Instruct base using LoRA (Low-Rank Adaptation). This model is explicitly designed as a defensive cybersecurity assistant, powering the PralayAI chatbot.

Key Capabilities

  • Incident Response: Provides step-by-step guidance for security events.
  • Threat Modeling: Assists with MITRE ATT&CK mapping and attack surface analysis.
  • Secure Coding: Helps identify and fix insecure code patterns.
  • Vulnerability Explanation: Explains OWASP Top 10, CVEs, and exploit concepts within a defensive context.
  • Safety Policy: Explicitly trained to refuse requests for offensive security tasks (e.g., phishing, malware creation, unauthorized exploitation) and instead offers safe, defensive alternatives.

Good For

  • Cybersecurity Education: Explaining complex concepts clearly for students and non-experts.
  • Security Analysts: Assisting with log analysis, malware defense strategies, and cloud security best practices.
  • Developers: Identifying and remediating secure coding vulnerabilities.

Limitations

As a 1.5B parameter model, it may have limitations in complex multi-step reasoning compared to larger models. Its knowledge cutoff means it lacks information on very recent CVEs or threat intelligence, and it is primarily trained on English content. It is intended as an assistant and not a replacement for professional security tools or certified analysts.

Finetunes
1 models