Search results

[Feature] Support FSDPEngine and RemoteSGLangEngine with factory classmethod

Status: Open.

#1168 In inclusionAI/AReaL;

· chenzhiyi021 opened on Apr 12, 2026

[Feature] Support heterogeneous cluster scheduling via Ray custom resources

Status: Open.

#1167 In inclusionAI/AReaL;

· dsuarezsoutoaws opened on Apr 12, 2026

[Doc] Docs/code alignment questions across config, workflows, LoRA, and metrics

documentationImprovements or additions to documentationImprovements or additions to documentation

Status: Open.

#1165 In inclusionAI/AReaL;

· KMing-L opened on Apr 12, 2026

[BUG] vLLM generation permanently paused after disk weight update, causing training to hang in Ascend NPU

bugSomething isn't workingSomething isn't working

Status: Open.

#1160 In inclusionAI/AReaL;

· chefwang-cloid opened on Apr 10, 2026

Post-#1157: redundant offload/onload transitions in train/save/eval/stats-export paths

call-for-contributionSomething planned but not in our current developmentSomething planned but not in our current development

good first issueGood for newcomersGood for newcomers

Status: Open.

#1158 In inclusionAI/AReaL;

· garrett4wade opened on Apr 9, 2026

[Question]Why keep warning "CLIArgs WARNING: behave_imp_weight_cap and behave_imp_weight_mode are configured but use_decoupled_loss=False. " even though I set use_decoupled_loss=True

call-for-contributionSomething planned but not in our current developmentSomething planned but not in our current development

good first issueGood for newcomersGood for newcomers

questionFurther information is requestedFurther information is requested

Status: Open.

#1156 In inclusionAI/AReaL;

· leekum2018 opened on Apr 9, 2026

[Question]Recommendation of versions of transformers and SGlang for running Qwen3.5 in Areal

bugSomething isn't workingSomething isn't working

call-for-contributionSomething planned but not in our current developmentSomething planned but not in our current development

good first issueGood for newcomersGood for newcomers

questionFurther information is requestedFurther information is requested

Status: Open.

#1155 In inclusionAI/AReaL;

· leekum2018 opened on Apr 9, 2026

[Feature] Support multi-node inference workers in inference gateway

call-for-contributionSomething planned but not in our current developmentSomething planned but not in our current development

good first issueGood for newcomersGood for newcomers

Status: Open.

#1149 In inclusionAI/AReaL;

· garrett4wade opened on Apr 8, 2026

DPO algo and WeChat QR-code support

questionFurther information is requestedFurther information is requested

Status: Open.

#1137 In inclusionAI/AReaL;

· ZKadelzq opened on Apr 4, 2026

feat(engine): sparse delta compression for disk-based weight updates

Status: Open.

#1125 In inclusionAI/AReaL;

· rchardx opened on Apr 1, 2026

[Feature] RTensor optimization: caching and IPC backend

call-for-contributionSomething planned but not in our current developmentSomething planned but not in our current development

enhancementNew feature or requestNew feature or request

good first issueGood for newcomersGood for newcomers

Status: Open.

#1117 In inclusionAI/AReaL;

· garrett4wade opened on Mar 31, 2026

[Question] VLM training fails with JSON serialization errors after applying PR #1070 based on NPU

questionFurther information is requestedFurther information is requested

Status: Open.

#1116 In inclusionAI/AReaL;

· Tb223001 opened on Mar 31, 2026