Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Research paper Hugging Face

SLAI T-Rex Achieves 34.22% MFU on Ascend NPU for DeepSeek-V4 Post-Training

AI By Crimson AI Hugging Face Papers 23 July 2026 · 00:00 8 views
Share: X Telegram

A new hierarchical optimization framework enables full-parameter post-training of trillion-parameter MoE models on Ascend NPU SuperPOD, achieving 2.93x MFU improvement over baseline and a specialized OR model outperforming GPT-5.4-Mini.

SLAI T-Rex Achieves 34.22% MFU on Ascend NPU for DeepSeek-V4 Post-Training

Key points

Researchers from SLAI have introduced SLAI T-Rex, a full-stack optimization framework for post-training trillion-parameter Mixture-of-Experts (MoE) models on Ascend NPU SuperPOD clusters. The work addresses critical system-level challenges such as memory pressure, communication overhead, and inefficient kernel execution that arise when training large-scale models on non-GPU hardware.

The framework employs a hierarchical approach spanning model-level parallelism, computation-communication orchestration, and low-level kernel optimization. On the DeepSeek-V4 model family, SLAI T-Rex achieves 34.22% Model FLOPs Utilization (MFU), a 2.93x improvement over the open-source baseline recipe, while maintaining training stability.

Building on this infrastructure, the team developed a domain-specific workflow for Operations Research (OR) tasks. They created curated Continuous Pretraining (CPT) and Supervised Fine-Tuning (SFT) data pipelines, combining collected domain resources with solver-verified synthetic optimization documents. The resulting dataset includes 10,000 high-quality SFT samples across four task categories and three problem representations.

The specialized model, DeepSeek-V4-Flash-OR, achieves the highest average zero-shot Pass@1 score among evaluated models at 71.81%, outperforming GPT-5.4-Mini by 3.98 percentage points and the base DeepSeek-V4-Flash model by 11.27 percentage points. The model and code are publicly available on ModelScope and GitHub.

MetricValue
MFU (SLAI T-Rex)34.22%
MFU improvement vs baseline2.93x
Zero-shot Pass@1 (DeepSeek-V4-Flash-OR)71.81%
Improvement vs GPT-5.4-Mini+3.98 pp
Improvement vs base DeepSeek-V4-Flash+11.27 pp
SFT samples10,000
Source
Hugging Face · Hugging Face Papers
Related news
Research paper
Hugging Face 31 Aug 2026

Hugging Face Unveils StepGuard: Step-Level Guardrails for Safer AI Agents

StepGuard, a new step-level guard model from Hugging Face, audits agent actions before execution, reducing attack success rates by...

1
Research paper
Hugging Face 31 Aug 2026

Hugging Face Researchers Unveil ABot-Recon for Stable Long-Horizon 3D Reconstruction

ABot-Recon, a new streaming 3D reconstruction model from Hugging Face, achieves stable long-horizon performance using only local t...

1
Research paper
Hugging Face 31 Aug 2026

ContextPilot: Teaching Agents Proactive Context Management via Fine-Grained RL

Hugging Face researchers introduce ContextPilot, a framework that enhances long-horizon agent reasoning by expanding context-editi...

1