Skip to content

Actions: flagos-ai/Megatron-LM-FL

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
2,500+ workflow runs
2,500+ workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
Auto Reminder Bot Auto Reminder Bot #229: Scheduled
1s main
main
1s
Approve Test Queue Approve Test Queue #5173: Scheduled
7s main
main
7s
fix(platform): platformize hardcoded CUDA references for v0.18.2 (MUSA) Request NVSkills CI #33: Issue comment #171 (comment) created by wbavon
1s
1s
fix(platform): platformize hardcoded CUDA references for v0.18.2 (MUSA) Claude Copy PR to Main #69: Issue comment #171 (comment) created by wbavon
1s
1s
fix(platform): platformize hardcoded CUDA references for v0.18.2 (MUSA) Claude Code Review #69: Issue comment #171 (comment) created by wbavon
1s
1s
Fix: use float32 for startup timestamp sync (ECCL rejects fp64/int64 all_reduce) Request NVSkills CI #32: Issue comment #170 (comment) created by wbavon
Startup failure
Startup failure
Fix: use float32 for startup timestamp sync (ECCL rejects fp64/int64 all_reduce) Claude Code Review #68: Issue comment #170 (comment) created by wbavon
1s
1s
Fix: use float32 for startup timestamp sync (ECCL rejects fp64/int64 all_reduce) Claude Copy PR to Main #68: Issue comment #170 (comment) created by wbavon
1s
1s
Fix: short-circuit single-rank param-group alignment to avoid all_gather_object garbage sizes Request NVSkills CI #31: Issue comment #169 (comment) created by wbavon
3s
3s
Fix: short-circuit single-rank param-group alignment to avoid all_gather_object garbage sizes Claude Copy PR to Main #67: Issue comment #169 (comment) created by wbavon
4s
4s
Fix: short-circuit single-rank param-group alignment to avoid all_gather_object garbage sizes Claude Code Review #67: Issue comment #169 (comment) created by wbavon
1s
1s
fix(training): platformize raw torch.cuda calls for non-CUDA backends Request NVSkills CI #30: Issue comment #168 (comment) created by wbavon
1s
1s
fix(training): platformize raw torch.cuda calls for non-CUDA backends Claude Code Review #66: Issue comment #168 (comment) created by wbavon
4s
4s
fix(training): platformize raw torch.cuda calls for non-CUDA backends Claude Copy PR to Main #66: Issue comment #168 (comment) created by wbavon
7s
7s
feat(platform): add Cambricon MLU platform registration Request NVSkills CI #29: Issue comment #167 (comment) created by wbavon
Startup failure
Startup failure
feat(platform): add Cambricon MLU platform registration Claude Code Review #65: Issue comment #167 (comment) created by wbavon
1s
1s
feat(platform): add Cambricon MLU platform registration Claude Copy PR to Main #65: Issue comment #167 (comment) created by wbavon
1s
1s
fix(core): degrade gracefully when persist_layer_norm is unsupported Request NVSkills CI #28: Issue comment #166 (comment) created by wbavon
Startup failure
Startup failure
fix(core): degrade gracefully when persist_layer_norm is unsupported Claude Copy PR to Main #64: Issue comment #166 (comment) created by wbavon
6s
6s
fix(core): degrade gracefully when persist_layer_norm is unsupported Claude Code Review #64: Issue comment #166 (comment) created by wbavon
1s
1s
fix(core): make jit_fuser a lazy decorator so --disable-jit-fuser takes effect Request NVSkills CI #27: Issue comment #165 (comment) created by wbavon
1s
1s
fix(core): make jit_fuser a lazy decorator so --disable-jit-fuser takes effect Claude Code Review #63: Issue comment #165 (comment) created by wbavon
8s
8s
fix(core): make jit_fuser a lazy decorator so --disable-jit-fuser takes effect Claude Copy PR to Main #63: Issue comment #165 (comment) created by wbavon
2s
2s
feat(npu): paged attention + platform convergence for Ascend NPU inference/RL Request NVSkills CI #26: Issue comment #164 (comment) created by wbavon
Startup failure
Startup failure
feat(npu): paged attention + platform convergence for Ascend NPU inference/RL Claude Code Review #62: Issue comment #164 (comment) created by wbavon
2s
2s