sealad886/FuseO1-DeepSeekR1-Qwen2.5-Coder-14B-Preview

TEXT GENERATIONPricing:Input $0.431 / Cached $0.0862 / Output $1.12Concurrent Unit Cost:1Model Size:14.8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Mar 17, 2025Architecture:Transformer Featherless Exclusive Cold

sealad886/FuseO1-DeepSeekR1-Qwen2.5-Coder-14B-Preview is a 14.8 billion parameter language model created by sealad886 through a merge of Qwen2.5-Coder-14B-Instruct and DeepSeek-R1-Distill-Qwen-14B using the sce merge method. This model is based on the Qwen2.5-Coder-14B architecture and is currently a work in progress. Its primary characteristic is its merged composition, aiming to combine strengths from its constituent models, particularly in coding-related tasks given its base.

Loading preview...

Overview

This model, FuseO1-DeepSeekR1-Qwen2.5-Coder-14B-Preview, is a 14.8 billion parameter language model developed by sealad886. It is a merged model, created using the mergekit tool, and is currently under active development, meaning its output may not yet be fully stable or valid.

Key Capabilities

  • Merged Architecture: Combines the strengths of Qwen/Qwen2.5-Coder-14B-Instruct and deepseek-ai/DeepSeek-R1-Distill-Qwen-14B.
  • Coder Base: Built upon the Qwen/Qwen2.5-Coder-14B architecture, suggesting a focus on code-related tasks.
  • Experimental: Represents an ongoing effort to explore model merging techniques for enhanced performance.

Good for

  • Experimentation: Ideal for researchers and developers interested in testing and contributing to models created via merging.
  • Code-centric Applications: Given its base models, it is likely intended for tasks involving code generation, understanding, or completion.
  • Exploring Mergekit: Provides a practical example of a model created using the sce merge method with mergekit.

Finetunes
1 models