nkasmanoff/qwen3.6-27b-opencode-lora-merged
nkasmanoff/qwen3.6-27b-opencode-lora-merged is a 27 billion parameter Qwen3.6 model, developed by nkasmanoff, fine-tuned with a LoRA adapter for opencode coding assistant tasks. It features a hybrid attention architecture and a substantial 262,144 token context length, supporting both text and vision modalities. This merged model is specifically optimized for AI-assisted coding, providing enhanced performance for developers using opencode.
Loading preview...
Model Overview
This model, nkasmanoff/qwen3.6-27b-opencode-lora-merged, is a specialized version of the Qwen3.6-27B base model, developed by nkasmanoff. It has been fine-tuned using a LoRA adapter specifically for opencode coding assistant tasks, with the LoRA weights merged directly into the base model for streamlined inference.
Key Capabilities & Features
- Base Model: Qwen/Qwen3.6-27B (27 billion parameters, BF16 precision).
- Architecture: Utilizes the Qwen3.5 architecture, featuring a hybrid attention mechanism that combines Mamba-2 linear attention with full attention every four layers.
- Context Length: Boasts an extensive context window of 262,144 tokens, enabling processing of very long codebases or complex prompts.
- Multimodal: Supports both text and vision inputs, making it versatile for various coding-related applications.
- Optimized for Opencode: Fine-tuned on opencode-specific data to enhance performance in AI-assisted coding workflows.
When to Use This Model
- AI-Assisted Coding: Ideal for developers seeking an advanced coding assistant, particularly when integrated with the opencode platform.
- Long Context Tasks: Excellent for scenarios requiring understanding and generation based on very large code files or extensive documentation.
- Multimodal Coding: Suitable for tasks that might involve both code and visual information, leveraging its text and vision capabilities.