grimjim/kuno-kunoichi-v1-DPO-v2-SLERP-7B

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Mar 9, 2024License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Cold

grimjim/kuno-kunoichi-v1-DPO-v2-SLERP-7B is a 7 billion parameter language model created by grimjim through a SLERP merge of SanjiWatsuki/Kunoichi-7B and SanjiWatsuki/Kunoichi-DPO-v2-7B. This model aims for increased robustness against errors and improved reasoning by combining two distinct base models. It supports ChatML and Alpaca prompting formats, with a context length of 4096 tokens.

Loading preview...

Model Overview

kuno-kunoichi-v1-DPO-v2-SLERP-7B is a 7 billion parameter language model developed by grimjim. It was created using the SLERP merge method via mergekit, combining two distinct base models: SanjiWatsuki/Kunoichi-7B and SanjiWatsuki/Kunoichi-DPO-v2-7B. The intention behind this merge is to enhance the model's resilience to errors and potentially improve its reasoning capabilities by integrating different implementations of comparable reasoning.

Key Features & Capabilities

  • Merge-based Architecture: Leverages the SLERP method to combine two Kunoichi models, aiming for a more robust and potentially diverse reasoning foundation.
  • Prompt Format Support: Compatible with both ChatML and Alpaca prompting formats, offering flexibility for integration into various applications.
  • Context Length: Supports a context window of 4096 tokens.
  • Quantization Availability: GGUF and GGUF-IQ-Imatrix quantizations are available, provided by Lewdiculous, for optimized inference on different hardware.

Intended Use Cases

This model is suitable for general language generation tasks where a 7B parameter model is appropriate. Its merged nature suggests potential benefits in scenarios requiring robust and varied responses, particularly in conversational or instruction-following applications given its support for ChatML and Alpaca formats.