KAT-Dev-72B-Exp-GGUF

GGUF quants of Kwaipilot/KAT-Dev-72B-Exp

Using llama.cpp b6730 (commit e60f01d941bc5b7fae62dd57fee4cec76ec0ea6e)

The importance matrix was generated with calibration_datav3.txt.

All quants were generated/calibrated with the imatrix, including the K quants.

---

<div align="center"> <img src="https://cdn-uploads.huggingface.co/production/uploads/61ee40a269351366e29972ad/KIYEa1c_WJEWPpeS0L_k1.png" width="100%" alt="Kwaipilot" /> </div> <hr>

News

🔥 We’re thrilled to announce the release of KAT-Dev-72B-Exp, our latest and most powerful model yet!

🔥 You can now try our strongest proprietary coder model KAT-Coder directly on the StreamLake platform for free.

Highlights

KAT-Dev-72B-Exp is an open-source 72B-parameter model for software engineering tasks.

On SWE-Bench Verified, KAT-Dev-72B-Exp achieves 74.6% accuracy ⚡ — when evaluated strictly with the SWE-agent scaffold.

KAT-Dev-72B-Exp is the experimental reinforcement-learning version of the KAT-Coder model. Through this open-source release, we aim to reveal the technical innovations behind KAT-Coder’s large-scale RL to developers and researchers.

Introduction

We rewrote the attention kernel and redesigned the training engine for shared prefix trajectories to achieve highly efficient RL training, especially for scaffolds leveraging context management.

Furthermore, to prevent exploration collapse observed in RL training, we reshaped advantage distribution based on pass rates: amplifying the advantage scale of highly exploratory groups while reducing that of low-exploration ones.

SWE agent Evaluation Parameters

temperature: 0.6
max_turns: 150
history_processors.n: 100

For full settings please refer to inference.yaml