Neph0s/CoSER-Llama-3.1-8B

Hugging Face
TEXT GENERATIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Feb 25, 2025License:mitArchitecture:Transformer0.0K Open Weights Featherless Exclusive Warm

Neph0s/CoSER-Llama-3.1-8B is an 8 billion parameter language model fine-tuned from LLaMA-3.1, specifically designed for role-playing language agents (RPLAs). Trained on the CoSER dataset of authentic multi-turn, multi-character dialogues from 771 novels, it excels at generating highly human-like responses across diverse personas. This model is optimized for capturing nuanced personalities and maintaining consistent character traits in complex role-playing scenarios, outperforming other open-source LLMs on RPLA benchmarks.

Loading preview...

CoSER-Llama-3.1-8B: Advanced Role-Playing Language Agent

Neph0s/CoSER-Llama-3.1-8B is an 8 billion parameter model built upon the LLaMA-3.1 base, specifically fine-tuned for role-playing language agents (RPLAs). It leverages the unique CoSER dataset, which comprises authentic multi-turn, multi-character dialogues extracted from 771 renowned novels, ensuring high fidelity and complexity in its training data.

Key Capabilities

  • Exceptional Role-Playing: Generates highly human-like responses, capturing nuanced personalities and maintaining consistent character traits for both established fictional and original characters.
  • Authentic Dialogue Generation: Trained on a dataset rich with character profiles, dialogues, plot summaries, and internal thoughts, enabling it to produce complex and realistic conversations.
  • State-of-the-Art Performance: Outperforms other open-source LLMs on multiple RPLA benchmarks, demonstrating strong capabilities in storyline consistency, anthropomorphism, and character fidelity.
  • Given-Circumstance Acting (GCA) Training: Utilizes a novel training approach where the model sequentially portrays characters within a given setting, optimizing for language modeling loss on each character's messages.

Should You Use This Model?

This model is ideal for applications requiring sophisticated character interaction and narrative generation. Consider CoSER-Llama-3.1-8B if your use case involves:

  • Interactive Storytelling: Creating dynamic and engaging narratives with consistent character voices.
  • Virtual Companions/Chatbots: Developing agents that can adopt and maintain specific personas.
  • Game Development: Populating game worlds with characters capable of realistic and context-aware dialogue.
  • Creative Writing Assistance: Generating dialogue or character interactions for literary projects.

It is particularly suited for scenarios where deep character immersion and consistent persona maintenance are critical, offering a significant advantage over general-purpose LLMs for role-playing tasks.