AITOPAI热报
ProductsOfficial

Prime Intellect 发布 Prime Mega MoE：让 Mega MoE 内核支持训练

Prime Intellect Releases Prime Mega MoE: Trainable Fused MoE Kernel, 3.11x Faster Forward Pass

官方一手marktechpost@Asif Razzaq
Blog/Media
Read the original
Why we picked it

Prime Intellect 把 DeepSeek 的 Mega MoE 内核加上了反向传播，Qwen3 MoE 训练能快 34%，做 MoE 训练的可以看看。

Prime Intellect 发布 Prime Mega MoE，为 DeepSeek 的融合 Mega MoE 内核补上了反向传播，使该内核可用于训练。在 prime-rl 框架中，Qwen3 MoE 的训练步骤最高提速 34%。该内核的前向传播速度达原版的 3.11 倍。

Image · marktechpost
Original · marktechpost

Prime Intellect Releases Prime Mega MoE: Trainable Fused MoE Kernel, 3.11x Faster Forward Pass

Prime Intellect released Prime Mega MoE, a backward pass for DeepSeek’s fused Mega MoE kernel. It makes the kernel trainable and speeds Qwen3 MoE training steps by up to 34% in prime-rl. The post Prime Intellect Releases Prime Mega MoE: Trainable Fused MoE Kernel, 3.11x Faster Forward Pass appeared first on MarkTechPost .

#Prime Intellect#Mega MoE#DeepSeek#Qwen3#MoE训练

Keep reading

Recent stories with the same tag or section
ProductsBuildersMulti-sourceSelected04:18
Prime Intellect launches Prime Inference serving GLM-5.3 on vLLM

Serving agentic workloads reliably is a systems problem beyond raw token throughput, and this launch shows how topology and scheduler design affect tool-call reliability. Practitioners building agent infrastructure can study the approach or evaluate Prime Inference against self-hosted vLLM deployments. GLM-5.3, a Zhipu (Z.ai) model, being served at scale also signals growing demand for open Chinese models in Western inference stacks.

#Prime Intellect#GLM-5.3#vLLM#agents
Event
官方账号vLLM@vllm_project5 sourcesSource