@modal: See how @ProximalHQ built a custom post-training stack with Modal Clusters for on-demand multi-node clusters, and autos…
Summary
Proximal 分享了其基于 Modal Clusters 构建的自研后训练基础设施，实现按需多节点集群与 GPU 自动扩缩容，以应对突发推理负载并打造高质量训练数据。
View Cached Full Text
Cached at: 10/08/26, 10:06 PM
See how @ProximalHQ built a custom post-training stack with Modal Clusters for on-demand multi-node clusters, and autoscaling GPUs to handle bursty inference workloads.
Proximal (@ProximalHQ): At Proximal, we believe that building our own research infrastructure is crucial to design the highest-quality training data
Today, we are sharing more information about our internal post-training stack
Similar Articles
@modal: The @datologyai team tested their Curation Studio at scale by generating large amounts of synthetic data on Modal, scal…
DatologyAI 发布 Curation Studio，面向训练自研模型的团队提供可定制的前沿数据筛选服务，并在 Modal 上利用数百块 B300 GPU 进行大规模合成数据测试验证。
@modal: Tinker changed how people do post-training. We're big fans, and so are many of the teams we work with. But as those tea…
Modal 发布 Spindle，一个开源、与 Tinker 兼容的训练服务器，可在 Modal 工作区中运行，让现有 Tinker 脚本无需改动即可迁移，并以 GPU 时间计费取代按 token 计费。
@quasagroup: Modal Review: Run AI Workloads Without Managing Servers https://youtu.be/a8vuwYK90-Q?si=vYj7YpobSXGNCDpQ… via @YouTube
Modal is a serverless cloud platform designed for AI workloads, supporting inference, training, and sandboxes. It enables instant scaling from zero to thousands of GPUs with pure Python code, significantly reducing latency and accelerating time to market.
@modal: Just launched at Modal Runtime: - Multi-node clusters - VM Sandboxes - Sandbox Sidecars - Sticky Sessions And more:
Modal announced several new capabilities at Modal Runtime, including multi-node clusters, VM Sandboxes, Sandbox Sidecars, and Sticky Sessions, expanding its platform for scalable AI and compute workloads.
@charles_irl: Inference isn't everything, but it does require a new stack -- not Kubernetes, not SLURM. At @modal, we dove deep to bu…
Modal engineers detail their approach to achieving truly serverless GPUs for AI inference, combining cloud buffers, a custom content-addressed filesystem, and CPU/GPU checkpoint/restore to scale replicas in tens of seconds instead of minutes.