nbeerbower/Qwen3-8B-abliterated-TIES
nbeerbower/Qwen3-8B-abliterated-TIES is an 8 billion parameter language model based on the Qwen3 architecture, created by nbeerbower through a TIES merge. This model integrates huihui-ai/Qwen3-8B-abliterated with Qwen/Qwen3-8B-Base, leveraging the TIES merging method to combine their strengths. It is designed for general language tasks, offering a 32768 token context length for processing extensive inputs.
Loading preview...
Model Overview
nbeerbower/Qwen3-8B-abliterated-TIES is an 8 billion parameter language model developed by nbeerbower. It is a merged model, combining the strengths of existing pre-trained language models using the TIES (Trimmed, Injected, and Scaled) merging method.
Merge Details
This model was constructed using mergekit with Qwen/Qwen3-8B-Base as its foundational base model. The primary model integrated into this merge is huihui-ai/Qwen3-8B-abliterated. The TIES method was applied with specific configurations, including a weight and density of 1 for the merged model, and normalization enabled. The merge process also utilized int8_mask and was performed with bfloat16 data type.
Key Characteristics
- Architecture: Based on the Qwen3 family of models.
- Parameter Count: 8 billion parameters, offering a balance between performance and computational efficiency.
- Context Length: Supports a substantial context window of 32768 tokens, suitable for handling longer texts and complex queries.
- Merging Method: Utilizes the TIES method, which is designed to effectively combine the knowledge and capabilities of multiple source models into a single, more robust model.