Search results

ActorDiedError during GRPO training with train_mcore_sglang_qwen3_next_grpo.sh (Ray worker exits with SYSTEM_ERROR / connection error code 2)

Status: Open.

#392 In alibaba/ChatLearn;

· yasu-nishi opened on Oct 29, 2025

TypeError: argument 'ids': 'list' object cannot be interpreted as an integer

Status: Open.

#358 In alibaba/ChatLearn;

· LOVE-YOURSELF-1 opened on Sep 2, 2025

[BUG] Resource temporarily unavailable

Status: Open.

#262 In alibaba/ChatLearn;

· lxy-nlp opened on Feb 26, 2025

[BUG] env var ENABLE_VLLM is not natively a bool

Status: Open.

#261 In alibaba/ChatLearn;

· slimfrkha opened on Feb 14, 2025

[BUG] fails to utilize all instances when num_replica of next model is less than the number of output batch_size

Status: Open.

#209 In alibaba/ChatLearn;

· haolin-nju opened on Jan 17, 2025

megatron_utils.py in load_checkpoint: args.iteration = state_dict['iteration'] TypeError: 'NoneType' object is not subscriptable

Status: Open.

#198 In alibaba/ChatLearn;

· carrot0117 opened on Jan 2, 2025

[train_rlhf_llama] When using vllm，Time-consuming optimization is not obvious as expectation.

Status: Open.

#197 In alibaba/ChatLearn;

· tkqie opened on Dec 31, 2024

[BUG] Context parallelism not enabled in SFT

Status: Open.

#189 In alibaba/ChatLearn;

· slimfrkha opened on Dec 24, 2024

[Alignment training (DPO)] Megatron support for QWEN2.5 series models

Status: Open.

#183 In alibaba/ChatLearn;

· Hevans123 opened on Dec 18, 2024

[Feature]Use Megatron-core dist_checkpointing to load checkpoint with different parallel strategies

Status: Open.

#169 In alibaba/ChatLearn;

· SeaOfOcean opened on Dec 5, 2024

[Feature]support to parameter sync when trainer_tp divides inference_tp for megatron core model

Status: Open.

#105 In alibaba/ChatLearn;

· charles9304 opened on Sep 25, 2024