v000000/L3.1-Niitorm-8B-t0.0001
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:Aug 28, 2024Architecture:Transformer0.0K Cold

L3.1-Niitorm-8B-t0.0001 is an 8 billion parameter language model based on the Llama 3.1 architecture, created by v000000. This model is a merge of Sao10K/L3.1-8B-Niitama-v1.1 and akjindal53244/Llama-3.1-Storm-8B, utilizing the NEARSWAP algorithm with a t0.0001 parameter. It is specifically designed as an RP (Role Play) model, leveraging the strengths of its merged components for enhanced interactive narrative generation. The model has a context length of 32768 tokens.

Loading preview...

Overview

L3.1-Niitorm-8B-t0.0001 is an 8 billion parameter language model developed by v000000, built upon the Llama 3.1 architecture. It is a product of a sophisticated merge operation, combining the base model Sao10K/L3.1-8B-Niitama-v1.1 (which includes grimjim/Llama-3-Instruct-abliteration-LoRA-8B) with akjindal53244/Llama-3.1-Storm-8B.

Merge Details

This model was created using the NEARSWAP merge algorithm with a t parameter of 0.0001. The merge process involved specific layer ranges from both source models, aiming to integrate their respective strengths. The base model for the merge was Sao10K/L3.1-8B-Niitama-v1.1+grimjim/Llama-3-Instruct-abliteration-LoRA-8B.

Key Capabilities

  • Role Play (RP) Optimization: The model is explicitly designed and optimized for role-playing scenarios, indicating a focus on generating coherent and engaging narrative responses in interactive contexts.
  • Llama 3.1 Foundation: Benefits from the underlying architecture and capabilities of the Llama 3.1 series, providing a strong base for language understanding and generation.
  • Merged Intelligence: Combines elements from "Niitama" and "Storm" models, suggesting an attempt to blend different strengths, potentially in creativity and reasoning, for its RP focus.

Prompt Template

The model utilizes a specific prompt template for interaction:

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

{system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>

{input}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

{output}<|eot_id|>

Good For

  • Applications requiring advanced role-playing capabilities and interactive storytelling.
  • Developers looking for a Llama 3.1-based model with a specific focus on creative and dynamic conversational generation.

Quantized versions (GGUF) are available from mradermacher and QuantFactory for efficient deployment.