l3utterfly/mistral-7b-v0.1-layla-v3

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kPublished:Feb 23, 2024License:apache-2.0Architecture:Transformer Open Weights Cold

l3utterfly/mistral-7b-v0.1-layla-v3 is a 7 billion parameter Mistral-based language model developed by l3utterfly and funded by Layla Network. Fine-tuned on the OpenHermes 2.5 dataset, it is optimized for multi-turn conversational interactions and character impersonation. This model excels in generating coherent and contextually relevant responses for dynamic dialogue scenarios.

Loading preview...

Model Overview

l3utterfly/mistral-7b-v0.1-layla-v3 is a 7 billion parameter Mistral-based language model, developed by l3utterfly and funded by Layla Network. It is fine-tuned from the Mistral 7B model using the OpenHermes 2.5 dataset, with a focus on enhancing conversational abilities. The training data underwent specific preprocessing, including the removal of refusals and mentions of AI assistants, and the splitting of multi-turn dialogues into distinct conversation records.

Key Capabilities

  • Multi-turn Conversation: Optimized for engaging in extended, dynamic dialogues.
  • Character Impersonation: Designed to effectively adopt and maintain specific character personas.
  • Refusal-Free Responses: Pre-processing of the training data aimed to eliminate refusal-based outputs.

Good For

  • Interactive Chatbots: Ideal for applications requiring natural and flowing multi-turn conversations.
  • Role-playing Scenarios: Suitable for generating responses consistent with specific character roles.
  • Offline Personal Assistants: This model serves as the base for Layla, an offline personal assistant developed by Layla Network.

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