PS4Research/qa-sft-phi4-reasoning

TEXT GENERATIONConcurrency Cost:1Model Size:14.7BQuant:FP8Ctx Length:32kPublished:May 16, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The PS4Research/qa-sft-phi4-reasoning model is a 14.7 billion parameter language model developed by PS4Research, fine-tuned from unsloth/phi-4-reasoning-unsloth-bnb-4bit. This model was trained using Unsloth and Huggingface's TRL library, emphasizing efficient training. It is designed for reasoning tasks, leveraging its base architecture and fine-tuning for enhanced performance in this domain. With a 32768 token context length, it supports extensive input for complex reasoning challenges.

Loading preview...

Overview

PS4Research/qa-sft-phi4-reasoning is a 14.7 billion parameter language model developed by PS4Research. It is fine-tuned from the unsloth/phi-4-reasoning-unsloth-bnb-4bit base model, indicating a focus on reasoning capabilities. The model was trained with the assistance of Unsloth and Huggingface's TRL library, which facilitated a 2x faster training process.

Key Capabilities

  • Reasoning Tasks: Inherits and enhances reasoning capabilities from its base model, making it suitable for logical inference and problem-solving.
  • Efficient Training: Benefits from Unsloth's optimizations, allowing for faster fine-tuning.
  • Extended Context: Features a 32768 token context length, enabling it to process and understand longer, more complex inputs relevant to reasoning.

Good For

  • Applications requiring robust logical reasoning.
  • Tasks that benefit from processing extensive contextual information.
  • Developers looking for a model fine-tuned for reasoning with an efficient training lineage.