PARTAGES-dev/Qwen3-4B-PDAPT-SLERP
PARTAGES-dev/Qwen3-4B-PDAPT-SLERP is a 4 billion parameter language model based on the Qwen3 architecture, created by PARTAGES-dev through a SLERP merge. This model combines Qwen/Qwen3-4B-Base with a specialized Qwen3-4B-Base-PARTAGES-v2-1440 variant. It is designed to leverage the strengths of its merged components, offering a balanced performance profile for general language tasks.
Loading preview...
Model Overview
PARTAGES-dev/Qwen3-4B-PDAPT-SLERP is a 4 billion parameter language model derived from the Qwen3 architecture. This model was constructed using the SLERP (Spherical Linear Interpolation) merge method, a technique known for smoothly combining the parameter spaces of different models.
Merge Details
The model is a composite of two base models:
- Qwen/Qwen3-4B-Base: The foundational Qwen3 model.
- Qwen3-4B-Base-PARTAGES-v2-1440: A specialized variant, likely fine-tuned or adapted for specific purposes by PARTAGES-dev.
The merge process involved combining these two models across all 36 layers, with a t parameter value of 0.5, indicating an equal weighting during the SLERP interpolation. The merge was performed using mergekit and utilized bfloat16 for parameter precision.
Key Characteristics
- Architecture: Qwen3-based, a robust and widely recognized LLM family.
- Parameter Count: 4 billion parameters, offering a balance between performance and computational efficiency.
- Merge Method: SLERP, which aims to create a coherent and well-performing model by interpolating between the capabilities of its constituents.
Potential Use Cases
This model is suitable for a variety of general-purpose language tasks where the combined strengths of its base models are beneficial. Developers looking for a Qwen3-based model with potentially enhanced or specialized capabilities from the PARTAGES-v2 variant may find this model useful for applications requiring text generation, summarization, or question answering.