Xenon1/Xenon-3

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kPublished:Feb 4, 2024License:apache-2.0Architecture:Transformer Open Weights Cold

Xenon1/Xenon-3 is a 7 billion parameter instruction-tuned language model based on the Mistral-7B-v0.1 architecture, fine-tuned on the Ultrafeedback dataset. It leverages techniques from the "Self-Rewarding Language Models" paper to enhance its conversational abilities. This model is optimized for generating coherent and contextually relevant responses in instruction-following scenarios, supporting a context length of 8192 tokens.

Loading preview...

Xenon-3: Instruction-Tuned Mistral-7B-v0.1

Xenon-3 is a 7 billion parameter language model developed by Xenon1, built upon the robust Mistral-7B-v0.1 architecture. This model has undergone instruction fine-tuning using the Ultrafeedback dataset, incorporating methodologies outlined in the Self-Rewarding Language Models paper.

Key Capabilities & Features

  • Instruction Following: Designed to accurately interpret and respond to user instructions, making it suitable for conversational AI and task-oriented dialogues.
  • Mistral-7B-v0.1 Base: Inherits architectural strengths like Grouped-Query Attention, Sliding-Window Attention, and a Byte-fallback BPE tokenizer, contributing to efficient processing and performance.
  • Context Length: Supports an 8192-token context window, allowing for more extended and complex interactions.
  • Chat Template Support: Utilizes a specific instruction format ([INST]...[/INST]) that is compatible with Hugging Face's apply_chat_template() for seamless integration into applications.

When to Use Xenon-3

  • Conversational Agents: Ideal for chatbots and virtual assistants requiring nuanced instruction adherence.
  • Content Generation: Can be used for generating text based on specific prompts or scenarios.
  • Research & Experimentation: A solid base for further fine-tuning or exploring self-rewarding learning techniques.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p