wvnvwn/llama2-7b-chat-lr5e-5-agnews-lr5e-5-cbresta

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:4kPublished:Jul 31, 2026Architecture:Transformer Featherless Exclusive Cold

This is a 7 billion parameter Llama 2 Chat model, created by wvnvwn, that has been merged using the linear merge method. It combines a fine-tuned Llama 2 Chat model for safety, another fine-tuned Llama 2 Chat model, and the base Meta Llama 2 7B Chat model. This merge aims to leverage the strengths of its constituent models, particularly focusing on aspects related to safety and chat-based applications.

Loading preview...

Model Overview

This model, developed by wvnvwn, is a 7 billion parameter language model based on the Llama 2 Chat architecture. It was created using the linear merge method via mergekit, combining three distinct Llama 2 Chat variants to potentially enhance its capabilities.

Merge Details

The model integrates the following components:

  • A Llama 2 7B Chat model specifically fine-tuned for safety (llama2_7b-chat-Safety-FT-lr5e-5).
  • Another fine-tuned Llama 2 7B Chat model (llama2-7b-chat-lr5e-5-ssft-cb).
  • The foundational Meta Llama 2 7B Chat HF model.

Each component was assigned specific weights during the merge process, with the safety-fine-tuned model having a weight of 1.0, the second fine-tuned model 0.3, and the base Llama 2 Chat model -0.3. This configuration suggests an intent to emphasize the safety-oriented fine-tuning while adjusting the influence of the base model.

Potential Use Cases

This merged model is likely suitable for chat-based applications where safety and adherence to specific fine-tuning objectives are important. Its architecture and merging strategy indicate a focus on refining conversational AI behavior.