Nabhos/Lehja-RP-Gemma-4-26B-A4B-v0.1
Nabhos/Lehja-RP-Gemma-4-26B-A4B-v0.1 is an experimental 26 billion parameter Gemma-4 model, fine-tuned by Nabhos, specifically targeting improved Hinglish (Hindi written in Latin characters) roleplay (RP) generation. This model focuses on reducing cliche and repetition by training only the lm_head. With a context length of 32768 tokens, it aims to enhance nuanced conversational interactions in Hinglish.
Loading preview...
Model Overview
Nabhos/Lehja-RP-Gemma-4-26B-A4B-v0.1 is an experimental 26 billion parameter model based on the Gemma-4 architecture, developed by Nabhos. This particular iteration is a fine-tune of TrevorJS/gemma-4-26B-A4B-it-uncensored, with a specific focus on training only the lm_head.
Key Characteristics
- Experimental Fine-tuning: The model is an experimental tune, primarily focusing on modifying the
lm_headto influence output generation. - Hinglish Roleplay (RP) Focus: It is specifically aimed at improving Hinglish (Hindi written in Latin characters) roleplay scenarios.
- Reduced Repetition: The primary goal of this fine-tuning approach is to mitigate cliche and reduce repetitive phrases in generated text.
- Context Length: The model supports a substantial context length of 32768 tokens, allowing for more extensive conversational turns.
Intended Use Cases
This model is currently in an experimental phase, and further testing is required to fully ascertain its capabilities. However, based on its design, it is intended for:
- Hinglish Conversational AI: Developing applications that require nuanced and less repetitive dialogue in Hinglish.
- Roleplay Scenarios: Generating creative and engaging roleplay content in Hinglish.
- Research and Development: Serving as a base for further experimentation in fine-tuning large language models for specific linguistic and stylistic improvements.