WonseokJayJung/cai-merge-YFlT4eBGlahNwO3waauW9k5Icf72-ai-ddr-1-0625-002324

TEXT GENERATIONConcurrent Unit Cost:1Model Size:0.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jun 25, 2026Architecture:Transformer Featherless Exclusive Cold

WonseokJayJung/cai-merge-YFlT4eBGlahNwO3waauW9k5Icf72-ai-ddr-1-0625-002324 is a 0.5 billion parameter language model created by WonseokJayJung, merged from Qwen2.5-Coder-0.5B-Instruct and Qwen2.5-0.5B using the SLERP method. This model combines the instruction-following and coding capabilities of the Coder variant with the general language understanding of the base Qwen2.5 model. It is designed for tasks requiring a balance of general text generation and code-related instruction processing within a compact 0.5B parameter footprint and a 32768 token context length.

Loading preview...

Overview

This model, created by WonseokJayJung, is a 0.5 billion parameter language model resulting from a merge of two Qwen 2.5 models: Qwen/Qwen2.5-Coder-0.5B-Instruct and Qwen/Qwen2.5-0.5B. The merge was performed using the SLERP (Spherical Linear Interpolation) method, aiming to combine the strengths of both base models.

Key Capabilities

  • Instruction Following: Inherits instruction-tuned capabilities from the Qwen2.5-Coder-0.5B-Instruct component.
  • Code Generation & Understanding: Benefits from the coding expertise of the Qwen2.5-Coder model, making it suitable for code-related tasks.
  • General Language Understanding: Retains the foundational language capabilities of the Qwen2.5-0.5B base model.
  • Compact Size: At 0.5 billion parameters, it offers a relatively small footprint for efficient deployment.
  • Extended Context: Supports a context length of 32768 tokens, allowing for processing longer inputs.

When to Use This Model

This merged model is particularly well-suited for use cases that require:

  • Resource-constrained environments where a larger model is not feasible.
  • Applications needing both general text generation and basic code-related instruction processing.
  • Tasks benefiting from a balance of instruction-following and foundational language understanding within a compact model size.