KeefeBuild/Keefe-Discere

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 3, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

KeefeBuild/Keefe-Discere is a 7.6 billion parameter Qwen2-based causal language model developed by KeefeBuild. This model was finetuned from KeefeBuild/Keefe-Discere-v3.0-Ultimate and optimized for training speed using Unsloth and Huggingface's TRL library. It features a 32768 token context length, making it suitable for applications requiring efficient processing of longer sequences.

Loading preview...

KeefeBuild/Keefe-Discere: An Efficiently Trained Qwen2 Model

KeefeBuild/Keefe-Discere is a 7.6 billion parameter language model developed by KeefeBuild. It is based on the Qwen2 architecture and was finetuned from the KeefeBuild/Keefe-Discere-v3.0-Ultimate model. A key highlight of this model is its training methodology, which leveraged Unsloth and Huggingface's TRL library to achieve significantly faster training times.

Key Characteristics

  • Architecture: Qwen2-based causal language model.
  • Parameter Count: 7.6 billion parameters.
  • Context Length: Supports a substantial context window of 32768 tokens.
  • Training Efficiency: Optimized for training speed using Unsloth, resulting in 2x faster finetuning.

Ideal Use Cases

This model is particularly well-suited for developers and researchers looking for:

  • Efficient Deployment: Models trained with optimized methods can offer better performance-to-resource ratios.
  • Applications requiring long context: The 32768 token context length supports complex tasks involving extensive text.
  • Further Finetuning: Its foundation and efficient training suggest it could be a strong base for additional domain-specific finetuning.