allura-org/MoE-Girl_400MA_1BT

TEXT GENERATIONConcurrent Unit Cost:1Model Size:1BQuant:BF16Context Size:32kPublished:Oct 22, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

MoE-Girl_400MA_1BT is a 1 billion parameter language model developed by allura-org, fine-tuned from IBM's Granite 3.0 architecture. This model is specifically designed and optimized for roleplaying scenarios, offering a specialized experience for interactive narrative generation. While primarily focused on roleplaying, it can also be adapted for general use cases, providing a versatile option for various text generation tasks.

Loading preview...

MoE Girl 400mA 1bT Overview

MoE-Girl_400MA_1BT is a specialized language model developed by allura-org, built upon IBM's Granite 3.0 architecture. With 400 million active parameters, this model is a compact yet capable option for specific applications. It has been meticulously fine-tuned with a "Special Sauce" of datasets, aiming to deliver a distinct user experience.

Key Capabilities

  • Roleplaying Focus: The primary design goal of MoE-Girl_400MA_1BT is to excel in roleplaying scenarios, making it suitable for interactive storytelling and character-driven conversations.
  • Granite 3.0 Base: Leverages the foundational strengths of IBM's Granite 3.0 model.
  • Compact Size: With 400 million active parameters, it offers a smaller footprint compared to larger models, potentially allowing for more efficient deployment in certain environments.

Good For

  • Roleplaying Applications: Ideal for developers and users looking to create engaging and dynamic roleplaying experiences.
  • Exploratory General Use: While optimized for roleplaying, it can be experimented with for general language generation tasks, though performance may be more akin to models like GPT-2 due to its parameter count.

Usage Notes

  • Prompting: The model is designed to be used with the ChatML format for optimal interaction. An example of the required format is provided in the original documentation.
  • Performance Expectations: Users should manage expectations regarding its performance, as its 400 million active parameters place it in a different league than much larger, more general-purpose LLMs.

Finetunes
3 models