Skip to content

Actions: NVIDIA/Megatron-LM

Actions

Community Bot

Actions

Loading...
Loading

Show workflow options

Create status badge

Loading
community-bot.yml
This workflow was disabled manually.
103 workflow runs
103 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

[QUESTION]NCCL timeout error when the second iteration Community Bot #103: Issue #1141 reopened by sbhavani
July 31, 2025 06:02 11s
July 31, 2025 06:02 11s
[QUESTION] Why does GPTDataset not directly cache all samples document_index and sample_index, and then construct different shuffle_index for different parameters? Community Bot #102: Issue #1139 reopened by sbhavani
July 31, 2025 04:36 12s
July 31, 2025 04:36 12s
[BUG] Learning rate not overrided when set --override-opt_param-scheduler Community Bot #101: Issue #1138 reopened by sbhavani
July 31, 2025 04:35 12s
July 31, 2025 04:35 12s
[QUESTION] Epochs Larger Than 1 When Specified with Trained Samples Community Bot #100: Issue #1127 reopened by sbhavani
July 31, 2025 04:35 13s
July 31, 2025 04:35 13s
Benchmarking DeepEP Guide Community Bot #99: Issue #1721 opened by sbhavani
July 31, 2025 03:59 11s
July 31, 2025 03:59 11s
[QUESTION]May I allocate and register NCCL userbuffer for activations/gradients transmitted between PP stages? Community Bot #98: Issue #1720 opened by yanminjia
July 31, 2025 01:35 9s
July 31, 2025 01:35 9s
How to split the num_layers unevenly when using pipeline parallelism? Community Bot #97: Issue comment #381 (comment) created by edgeinfinity1
July 30, 2025 14:54 16s
July 30, 2025 14:54 16s
[QUESTION] save_checkpoint with expert_tensor_parallel_size Community Bot #96: Issue #1719 edited by jeromeku
July 30, 2025 14:15 10s
July 30, 2025 14:15 10s
[QUESTION] save_checkpoint with expert_tensor_parallel_size Community Bot #95: Issue #1719 opened by jeromeku
July 30, 2025 14:14 12s
July 30, 2025 14:14 12s
v0.12.2 pretrain failed, 0.14.0rc2 model conversion failed Community Bot #94: Issue #1718 edited by justalittlenoob
July 30, 2025 09:32 12s
July 30, 2025 09:32 12s
v0.12.2 pretrain failed, 0.14.0rc2 model conversion failed Community Bot #93: Issue #1718 opened by justalittlenoob
July 30, 2025 08:10 10s
July 30, 2025 08:10 10s
[QUESTION]How to calculate MFU based on the flops? Community Bot #92: Issue #1565 closed by Lynnzake
July 30, 2025 06:44 13s
July 30, 2025 06:44 13s
[QUESTION] Purpose of all2all communication for probs in MoEAlltoAllTokenDispatcher Community Bot #91: Issue comment #1716 (comment) created by snowpeakz
July 30, 2025 06:22 11s
July 30, 2025 06:22 11s
[QUESTION] Purpose of all2all communication for probs in MoEAlltoAllTokenDispatcher Community Bot #90: Issue comment #1716 (comment) edited by yanring
July 30, 2025 02:26 9s
July 30, 2025 02:26 9s
[QUESTION] Purpose of all2all communication for probs in MoEAlltoAllTokenDispatcher Community Bot #89: Issue comment #1716 (comment) created by yanring
July 30, 2025 02:26 15s
July 30, 2025 02:26 15s
Issue 1672 fix: initializing the current pointed with int64 to avoid … Community Bot #88: Issue comment #1673 (comment) created by sharanmayank
July 29, 2025 23:33 11s
July 29, 2025 23:33 11s
[QUESTION] Performance Impact of Using item() in total_num_tokens += num_tokens.item() in megatron/core/pipeline_parallel/schedules.py Community Bot #87: Issue #1403 closed by sbhavani
July 29, 2025 22:08 12s
July 29, 2025 22:08 12s
[QUESTION]How to convert the weight file format of the MAMBA model from pt to safetensors format? Community Bot #86: Issue #1339 reopened by sbhavani
July 29, 2025 22:07 9s
July 29, 2025 22:07 9s
[BUG]异步保存参数async-save和断点续训之间存在bug Community Bot #85: Issue comment #1713 (comment) created by ananthsub
July 29, 2025 19:01 14s
July 29, 2025 19:01 14s
[QUESTION]Missing code branch Community Bot #84: Issue #1395 reopened by sbhavani
July 29, 2025 15:59 14s
July 29, 2025 15:59 14s
[BUG] can't load saved fp8 checkpoint when resume training Community Bot #83: Issue #1350 reopened by sbhavani
July 29, 2025 15:59 11s
July 29, 2025 15:59 11s
[QUESTION] Resume training about dataset Community Bot #82: Issue #1343 reopened by sbhavani
July 29, 2025 15:58 10s
July 29, 2025 15:58 10s
[BUG] Using different distributed strategies of Megatron-LM to train the llama3.1-8B model results in inconsistent training loss Community Bot #81: Issue #1324 reopened by sbhavani
July 29, 2025 15:56 9s
July 29, 2025 15:56 9s
[BUG] Zarr checkpoint loses distributed optimizer states due to lack of synchronizers on ranks that create arrays Community Bot #80: Issue comment #1053 (comment) created by sbhavani
July 29, 2025 15:53 12s
July 29, 2025 15:53 12s
[BUG] Zarr checkpoint loses distributed optimizer states due to lack of synchronizers on ranks that create arrays Community Bot #79: Issue #1053 closed by sbhavani
July 29, 2025 15:53 9s
July 29, 2025 15:53 9s