Nexusflow/Athene-70B-Preview

TEXT GENERATIONPricing:Input $3.5 / Cached $0.7 / Output $8.3Concurrent Unit Cost:4Model Size:70BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Jul 24, 2024License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Athene-70B is an open-weights 70 billion parameter chat model developed by the Nexusflow Team, fine-tuned from Llama-3-70B-Instruct using RLHF. This model is specifically optimized for conversational AI, achieving a high score of 77.8% on the Arena-Hard-Auto benchmark, positioning it competitively against proprietary models. It is designed for general-purpose chat applications where strong conversational performance is critical. The model utilizes an 8192 token context length, making it suitable for engaging and extended dialogue interactions.

Loading preview...

Athene-70B: An RLHF-Tuned Llama-3-70B Chat Model

Athene-70B is a 70 billion parameter open-weights large language model developed by the Nexusflow Team. It is built upon the robust Llama-3-70B-Instruct architecture and has undergone extensive Reinforcement Learning from Human Feedback (RLHF) to enhance its conversational capabilities.

Key Capabilities and Performance

  • High Conversational Performance: Athene-70B demonstrates strong performance in dialogue, achieving a notable score of 77.8% on the Arena-Hard-Auto benchmark. This places it competitively among leading proprietary models like Claude-3.5-Sonnet (79.3%) and GPT-4o (79.2%), significantly outperforming its base model, Llama-3-70B (46.6%).
  • RLHF Optimization: The model's training methodology focuses on RLHF, which is crucial for aligning model outputs with human preferences and improving chat quality.
  • Llama-3 Compatibility: It utilizes the same chat template as Llama-3-70B-Instruct, ensuring ease of integration for developers familiar with the Llama-3 ecosystem.

Use Cases and Considerations

Athene-70B is well-suited for applications requiring a high-performing, open-source chat model. Its strong benchmark results suggest it can be effectively deployed in:

  • General-purpose chatbots and conversational agents.
  • Interactive AI assistants.
  • Applications where robust dialogue generation and understanding are paramount.

Developers can leverage its Llama-3-compatible interface for straightforward implementation in existing workflows.

Finetunes
1 models