localized-ft/Llama-3.1-8B-target-only-no-hallucination-kld-seed4

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Aug 24, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The localized-ft/Llama-3.1-8B-target-only-no-hallucination-kld-seed4 is an 8 billion parameter Llama-3.1-based language model developed by localized-ft. This model was fine-tuned using Unsloth and Huggingface's TRL library, enabling 2x faster training. It is designed for specific target applications, focusing on reducing hallucination, and supports an 8192-token context length.

Loading preview...

Model Overview

localized-ft/Llama-3.1-8B-target-only-no-hallucination-kld-seed4 is an 8 billion parameter language model, fine-tuned by localized-ft. It is based on the Meta-Llama-3.1-8B-Instruct architecture and utilizes an 8192-token context window.

Key Characteristics

  • Efficient Fine-tuning: This model was fine-tuned using Unsloth and Huggingface's TRL library, resulting in a 2x speed improvement during the training process.
  • Targeted Optimization: The model is specifically designed to reduce hallucination, indicating a focus on factual accuracy and reliable output for particular use cases.

Potential Use Cases

  • Applications requiring high factual consistency and reduced generative errors.
  • Scenarios where efficient fine-tuning of Llama-3.1-based models is beneficial.
  • Tasks that can leverage an 8192-token context length for processing longer inputs.