longtermrisk/Llama-3.1-8B-school-of-reward-hacks-sft-seed3

TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Aug 15, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The longtermrisk/Llama-3.1-8B-school-of-reward-hacks-sft-seed3 is an 8 billion parameter Llama-3.1 instruction-tuned causal language model developed by longtermrisk. This model was fine-tuned using Unsloth and Huggingface's TRL library, enabling 2x faster training. It is designed for general language tasks, leveraging its Llama-3.1 architecture for broad applicability.

Loading preview...

Overview

This model, longtermrisk/Llama-3.1-8B-school-of-reward-hacks-sft-seed3, is an 8 billion parameter instruction-tuned language model. It is based on the Llama-3.1 architecture and was developed by longtermrisk.

Key Characteristics

  • Base Model: Fine-tuned from unsloth/Meta-Llama-3.1-8B-Instruct.
  • Training Efficiency: The fine-tuning process utilized Unsloth and Huggingface's TRL library, which allowed for a reported 2x faster training speed.
  • Context Length: Supports a context length of 8192 tokens.
  • License: Distributed under the Apache-2.0 license.

Good For

  • General Language Tasks: Suitable for a wide range of applications requiring a capable 8B parameter instruction-following model.
  • Efficient Fine-tuning Exploration: Demonstrates the potential for faster fine-tuning workflows using tools like Unsloth, which could be beneficial for developers looking to iterate quickly on Llama-3.1 models.