ljh728/kanana-1.5-8b-instruct-2505-Persona-Merged

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Sep 3, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The ljh728/kanana-1.5-8b-instruct-2505-Persona-Merged is an 8 billion parameter instruction-tuned Llama model developed by ljh728. This model was fine-tuned from kakaocorp/kanana-1.5-8b-instruct-2505 using Unsloth and Huggingface's TRL library, enabling 2x faster training. It is designed for general instruction-following tasks, leveraging its efficient training methodology.

Loading preview...

Model Overview

The ljh728/kanana-1.5-8b-instruct-2505-Persona-Merged is an 8 billion parameter instruction-tuned Llama model, developed by ljh728. It is based on the kakaocorp/kanana-1.5-8b-instruct-2505 model and was fine-tuned using a combination of Unsloth and Huggingface's TRL library. This approach facilitated a 2x faster training process compared to standard methods.

Key Characteristics

  • Base Model: Fine-tuned from kakaocorp/kanana-1.5-8b-instruct-2505.
  • Parameter Count: 8 billion parameters, offering a balance between performance and computational efficiency.
  • Training Efficiency: Utilizes Unsloth for accelerated training, making it a potentially resource-efficient option for deployment.
  • Context Length: Supports an 8192-token context window, suitable for handling moderately long inputs.

Potential Use Cases

This model is well-suited for general instruction-following applications where efficient inference and a solid understanding of prompts are required. Its optimized training suggests it could be a good candidate for scenarios needing a capable language model without excessive computational overhead.