Overview

This model, ertghiu256/Qwen3-4b-tcomanr-merge-v2.3, is a 4 billion parameter language model built upon the Qwen3 architecture. It is a merged model, created using the TIES method, with Qwen/Qwen3-4B-Thinking-2507 serving as its base. The primary goal of this merge was to significantly enhance the model's performance in reasoning, code generation, and mathematical problem-solving by integrating capabilities from various specialized Qwen3 fine-tunes.

Key Capabilities

• Enhanced Reasoning: Designed to handle complex reasoning tasks, building on its base model's strengths.
• Code Generation: Incorporates specific fine-tunes to improve its ability to generate and understand code.
• Mathematical Proficiency: Merges models focused on mathematical reasoning to boost performance in this area.
• Extended Context: Supports a substantial context length of 40960 tokens, allowing for processing longer inputs and maintaining coherence over extended interactions.

Merge Details

The model is a revision of ertghiu256/Qwen3-4b-tcomanr-merge-v2.2 and combines over a dozen different Qwen3-based models, including those specialized in thinking, code reasoning, math reasoning, and instruction following. This intricate merge process aims to create a versatile model proficient in multiple technical domains. The configuration used for the TIES merge is detailed in the original README, outlining the specific weights and densities applied to each contributing model.