prithivMLmods/Sombrero-R1-14B-Elite13

TEXT GENERATIONPricing:Input $0.431 / Cached $0.0862 / Output $1.12Concurrent Unit Cost:1Model Size:14.8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:May 23, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Sombrero-R1-14B-Elite13 by prithivMLmods is a 14.8 billion parameter language model, fine-tuned from DeepSeek-R1-Distill-Qwen-14B using reinforcement learning. It is optimized as a high-performance reasoning assistant, excelling in mathematical problem-solving and general-purpose conversational tasks. The model supports an expanded context of up to 128K tokens and demonstrates strong instruction adherence, making it suitable for complex interactive and educational applications.

Loading preview...

Sombrero-R1-14B-Elite13: A Reasoning-Optimized LLM

Sombrero-R1-14B-Elite13 is a 14.8 billion parameter model developed by prithivMLmods, based on the DeepSeek-R1-Distill-Qwen-14B architecture. This model has undergone significant enhancement through reinforcement learning fine-tuning, specifically to optimize its performance as a reasoning assistant.

Key Capabilities & Enhancements

  • Mathematical Reasoning Proficiency: Excels at solving mathematical problems, providing accurate solutions and step-by-step breakdowns across various domains like algebra, calculus, and logic puzzles.
  • Instruction Adherence: Demonstrates strong ability to understand and follow multi-part instructions and structured tasks.
  • Expanded Context Handling: Supports an impressive context length of up to 128K tokens, with output generation up to 8K tokens, making it suitable for processing extensive technical and educational content.
  • Cross-Domain Knowledge: Offers broad general knowledge, facilitating tutoring, research, and exploratory conversations.
  • Multilingual Support: Capable of assisting in over 29 languages, including English, Chinese, French, and German.

Intended Use Cases

This model is particularly well-suited for:

  • Mathematics Problem Solving: Ideal for derivations, symbolic computation, and numerical explanations.
  • Educational & Instructional Support: Provides guided explanations for students and instructors.
  • Chat-based Reasoning: Designed for coherent, context-aware dialogue with structured logic.
  • Document & Code Explanation: Can interpret and explain complex documents, code snippets, and logical flows.

Limitations

Users should be aware that the model is compute-intensive, requiring high-memory hardware (e.g., \u226548GB VRAM). Like many LLMs, it may exhibit potential for bias and hallucinations and can experience drift in long responses. Its knowledge is static, limited to its training data, and its performance in highly creative tasks may vary.

Finetunes
1 models