hungpill/kanana-1.5-8b-instruct-2505-Safe-DPO

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Sep 4, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The hungpill/kanana-1.5-8b-instruct-2505-Safe-DPO is an 8 billion parameter Llama-based instruction-tuned language model developed by hungpill. This model was fine-tuned using Unsloth and Huggingface's TRL library, enabling a 2x faster training process. It is designed for general instruction-following tasks, leveraging its efficient training methodology to provide a capable and accessible LLM solution.

Loading preview...

Model Overview

The hungpill/kanana-1.5-8b-instruct-2505-Safe-DPO is an 8 billion parameter instruction-tuned language model developed by hungpill. It is based on the Llama architecture and was fine-tuned from a previous hungpill model, hungpill/kanana-1.5-8b-instruct-2505-Safe-DPO.

Key Characteristics

  • Efficient Training: This model was trained significantly faster, specifically 2x faster, by utilizing the Unsloth library in conjunction with Huggingface's TRL library. This highlights an optimization in the training pipeline.
  • Instruction-Tuned: As an instruction-tuned model, it is designed to understand and follow user prompts effectively, making it suitable for a wide range of conversational and task-oriented applications.
  • Open License: The model is released under the Apache-2.0 license, promoting its use and further development within the community.

Use Cases

Given its instruction-tuned nature and efficient development, this model is well-suited for:

  • General-purpose conversational AI.
  • Text generation tasks requiring adherence to specific instructions.
  • Applications where a capable 8B parameter model with an optimized training history is beneficial.