trinhkhng/linear_Merged_Qwen2-0.5B_0.4

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 6, 2026Architecture:Transformer Featherless Exclusive Cold

The trinhkhng/linear_Merged_Qwen2-0.5B_0.4 is a 0.5 billion parameter language model created by trinhkhng, merged using the Linear method. This model combines a base Qwen2-0.5B with a debiased version of Qwen2-0.5B, with specific weighting applied during the merge. It is designed to leverage the strengths of both constituent models, potentially offering improved performance or reduced bias compared to the base Qwen2-0.5B.

Loading preview...

Model Overview

The trinhkhng/linear_Merged_Qwen2-0.5B_0.4 is a 0.5 billion parameter language model resulting from a merge operation. This model was constructed using the Linear merge method via mergekit, combining two distinct Qwen2-0.5B variants.

Merge Details

The merge process specifically combined a base /kaggle/working/Qwen2-0.5B model with a /kaggle/working/debias_Qwen2-0.5B model. A weighting configuration was applied, giving the base Qwen2-0.5B a weight of 0.6 and the debiased version a weight of 0.4, with normalization enabled. This approach aims to integrate the characteristics of both models into a single, cohesive unit.

Potential Use Cases

This merged model could be beneficial for applications where:

  • A compact 0.5 billion parameter model is required.
  • Leveraging the combined strengths of a base model and a debiased variant is advantageous.
  • Exploration of merged model performance for specific tasks is desired.