PulseAugur
EN
LIVE 03:27:13

Moore Threads open-sources trillion-parameter MoE training stack

By PulseAugur Editorial · [1 sources] ·

Moore Threads has released its Megatron-MUSA-Titan training stack as open-source. This stack is designed for training trillion-parameter Mixture-of-Experts (MoE) models and has been utilized on clusters of up to 10,000 GPUs. The system supports advanced features like blockwise FP8 precision and incorporates reinforcement learning techniques such as slime and miles RL. AI

IMPACT Enables researchers and developers to train extremely large MoE models more efficiently.

RANK_REASON Open-source release of a training stack for large-scale model training. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Pandaily →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Moore Threads open-sources trillion-parameter MoE training stack

How we ranked this

Signal score
43 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Open-source release of a training stack for large-scale model training. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. Pandaily TIER_1 English(EN) · [email protected] (Pandaily) ·

    Moore Threads Open-Sources Megatron-MUSA-Titan for Trillion-Parameter MoE Training on Its GPUs

    Moore Threads has open-sourced Megatron-MUSA-Titan, the training stack it uses for trillion-parameter MoE pretraining on 10,000-GPU clusters, with blockwise FP8 and support for slime and miles RL.

RELATED ENTITIES

RELATED TOPICS