Undi95/Lumimaid-Magnum-12B

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:12BQuant:FP8Ctx Length:32kPublished:Jul 30, 2024Architecture:Transformer0.0K Warm

Undi95/Lumimaid-Magnum-12B is a 12 billion parameter language model created by Undi95, resulting from a merge of the Lumimaid and Magnum models. This model incorporates a fine-tuned component based on Claude input, trained on a 16k context length, enhancing its conversational capabilities. It is optimized for generating responses in a Mistral-style prompt format, making it suitable for instruction-following tasks.

Loading preview...

Undi95/Lumimaid-Magnum-12B Overview

This model, developed by Undi95, is a 12 billion parameter language model created through a novel merge process using the DELLA method in mergekit. It combines the strengths of the Lumimaid and Magnum models.

Key Capabilities

  • Enhanced Conversational Ability: Integrates a fine-tuned component specifically trained on Claude input, suggesting improved performance in dialogue and instruction-following scenarios.
  • Extended Context Handling: The fine-tuned component was trained on a 16k context length, indicating a capacity for processing longer inputs and maintaining coherence over extended conversations.
  • Mistral Prompt Compatibility: Designed to work seamlessly with the Mistral prompt template, ensuring straightforward integration into existing workflows that utilize this format.

Good For

  • Applications requiring a model with a blend of Lumimaid and Magnum's characteristics.
  • Instruction-following tasks where a Mistral-style prompt is preferred.
  • Use cases benefiting from a model fine-tuned with Claude-like conversational patterns and a 16k context window.

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