All issues

Search results

【Error】 在cuda上跑简单的训练失败

Status: Open.

#1295 In flagos-ai/FlagScale;

· FatJhon opened on Sep 16, 2026

Tensor/context parallel training fails on non-CUDA platforms: CUDA_DEVICE_MAX_CONNECTIONS assertion

P2Priority 2 - mediumPriority 2 - medium

Status: Open.

Bug

#1289 In flagos-ai/FlagScale;

· kolinweiwei opened on Sep 8, 2026

VLA unit test collection fails on Kunlunxin P800 due to FlagGems and Triton API incompatibility

Status: Open.

#1270 In flagos-ai/FlagScale;

· wyNomad opened on Aug 17, 2026

FlagGems import fails on Kunlunxin P800 because current Triton lacks triton.language.math.asin

Status: Open.

#1265 In flagos-ai/FlagScale;

· wyNomad opened on Aug 13, 2026

是否支持异构芯片的混推

Status: Open.

#1217 In flagos-ai/FlagScale;

· wjiannan opened on Jun 9, 2026

[CICD]: The tokenizer used by cicd is deleted.

bugSomething isn't workingSomething isn't working

Status: Open.

Bug

#1206 In flagos-ai/FlagScale;

· LiJunscs opened on May 23, 2026

Bug of saving moe checkpoint when ckpt_format is torch

Status: Open.

#1188 In flagos-ai/FlagScale;

· LiJunscs opened on Apr 8, 2026

Bug of MoE checkpoint.

Status: Open.

#1185 In flagos-ai/FlagScale;

· LiJunscs opened on Apr 4, 2026

Use Jinja templating to generate runner scripts

enhancementNew feature or requestNew feature or request

Status: Open.

Feature

#1137 In flagos-ai/FlagScale;

· tengqm opened on Mar 6, 2026

How to start training task in slurm?

Status: Open.

#1136 In flagos-ai/FlagScale;

· LiJunscs opened on Mar 5, 2026

[Bug] Missing hydra-core in the installation requirements.

Status: Open.

#1097 In flagos-ai/FlagScale;

· lindylin1817 opened on Jan 30, 2026

功能类：「建议 FlagScale 增加启动参数，控制启动方式为前台运行还是后台运行」；

Status: Open.

#1090 In flagos-ai/FlagScale;

· lovywu opened on Jan 28, 2026