KaraKaraWitch/GoldDiamondGold-L33-70b

Warm
Public
70B
FP8
32768
Hugging Face
Overview

GoldDiamondGold-L33-70b Overview

KaraKaraWitch/GoldDiamondGold-L33-70b is a 70 billion parameter language model created by KaraKaraWitch using the SCE (Selective Component Ensemble) merge method. It is built on the deepcogito/cogito-v2-preview-llama-70B base model and incorporates elements from 17 other Llama-family models, including those focused on narration, creative writing, roleplaying, and tool calling.

Key Capabilities & Characteristics

  • Enhanced Intelligence: The model is noted for its improved intelligence compared to previous iterations and other models in its class, demonstrating a strong grasp of prompts and desired 'vibes'.
  • Nuanced Prose Style: It aims to combine the prose style and quality reminiscent of Claude 3 Sonnet/Opus, with specific components targeting better writing style and creativity for fiction.
  • Roleplaying & Story Writing: Several merged models contribute to its strengths in roleplaying and story generation, including enhancements for ERP/RP and anime-tilt dialogue.
  • System Prompt Sensitivity: The model is a "stickler for system prompts," performing optimally when provided with well-written and clear instructions.
  • Tool Calling: Includes components specifically for tool calling capabilities.

Intended Use Cases

This model is particularly well-suited for applications requiring:

  • Creative Writing: Generating nuanced narratives, fiction, and roleplay scenarios.
  • Intelligent Conversational Agents: Developing assistants that can understand and respond intelligently to complex prompts.
  • Structured Interactions: Leveraging its sensitivity to system prompts for controlled and predictable outputs.
  • Tool-Use Applications: Integrating with external tools through its tool-calling capabilities.