trinhkhng/nearswap_Merged_Qwen2-0.5B_0.3

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

trinhkhng/nearswap_Merged_Qwen2-0.5B_0.3 is a 0.5 billion parameter language model based on the Qwen2 architecture, created by trinhkhng. This model was produced using the NearSwap merge method, combining a base Qwen2-0.5B model with a debiased variant. It is designed for general language tasks, leveraging its merged architecture to potentially offer refined performance characteristics.

Loading preview...

Model Overview

trinhkhng/nearswap_Merged_Qwen2-0.5B_0.3 is a compact 0.5 billion parameter language model. It was developed by trinhkhng through a merging process using the NearSwap method, which combines pre-trained language models to create a new one.

Merge Details

This model's unique characteristic lies in its creation via mergekit. Specifically, it integrates a base /kaggle/working/Qwen2-0.5B model with a debiased version, /kaggle/working/debias_Qwen2-0.5B. The NearSwap method was applied with a t parameter of 0.3, indicating a specific weighting or configuration during the merge process. This approach aims to leverage the strengths of both constituent models.

Key Characteristics

  • Architecture: Based on the Qwen2 family.
  • Parameter Count: 0.5 billion parameters, making it suitable for applications requiring a smaller footprint.
  • Context Length: Supports a substantial context window of 32768 tokens.
  • Creation Method: Utilizes the NearSwap merging technique for model synthesis.

Potential Use Cases

Given its merged nature and smaller size, this model could be suitable for:

  • Resource-constrained environments: Where larger models are impractical.
  • Specific fine-tuning tasks: As a base for further specialization.
  • Exploration of merged model behaviors: For researchers interested in the effects of model merging strategies.