pytorch PiPPy issues - Githubissues

pytorch / PiPPy

Pipeline Parallelism for PyTorch

BSD 3-Clause "New" or "Revised" License

715 stars 86 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

How to train a model with pippy

#1142 sunkun1997 opened 2 weeks ago
0
fixed missing argument and refactoring

#1141 Ankur-singh opened 3 weeks ago
2
pippy.SaveModule not exist？

#1140 laoda513 opened 1 month ago
0
Update all hf examples to have dist.barrier

#1139 muellerzr closed 1 month ago
0
Code hangs permanently

#1138 Narasimha1997 opened 1 month ago
0
Fix llama example split failed

#1137 rednoah91 opened 1 month ago
1
Support for Autoregressive generation with LLMs

#1136 apresunreve opened 2 months ago
0
Meta init llama then pipeline then materialize

#1135 kwen2501 opened 2 months ago
1
[Error] pipeline() got an unexpected keyword argument

#1134 HieronZhang opened 2 months ago
1
[Bug?] Gradient Synchronization for DDP

#1133 jianweif opened 2 months ago
3
[BUG] cannot capture your model as a full graph

#1132 sunkun1997 opened 2 months ago
5
ModuleNotFoundError: No module named 'torch.distributed.pipelining'.

#1131 sunkun1997 closed 2 months ago
1
`pipeline` arguments are not matched

#1130 rednoah91 opened 2 months ago
8
Implemented flexible PP

#1129 haocizhang opened 3 months ago
1
Add migration notice

#1128 kwen2501 closed 3 months ago
0
Migrate Llama example to use torch APIs

#1127 kwen2501 closed 3 months ago
0
CPU offloading?

#1126 Xynonners opened 3 months ago
2
Move auto split out of GPT2 example into a separate file

#1125 kwen2501 closed 3 months ago
0
Migrate some of the HF examples to use 2.4 PP APIs

#1124 kwen2501 closed 3 months ago
0
ImportError: cannot import name 'pipeline' from 'pippy'

#1123 bob020416 closed 3 months ago
2
Can Pippy be combined with PEFT LoRA?

#1122 Songjw133 opened 3 months ago
1
Add nightly model tests against pytorch

#1121 kwen2501 closed 3 months ago
1
Add nightly model tests against pytorch

#1120 kwen2501 closed 4 months ago
0
Adding 'labels' input to model with 'include_loss_args' fails hf examples

#1119 alexlan137 opened 4 months ago
0
Inference freezes when running llama example with pp>2

#1118 JamesLYan opened 4 months ago
3
[WIP] enable doraPP

#1117 tianfengfrank opened 4 months ago
1
Refactor HuggingFace examples to use torch.distributed.pipelining

#1116 kwen2501 closed 4 months ago
0
examples/huggingface failed

#1115 yaxan opened 4 months ago
8
add ddp test

#1114 H-Huang closed 4 months ago
0
Privatize step_microbatches

#1113 kwen2501 closed 4 months ago
0
Update test_cpu_init

#1112 kwen2501 closed 4 months ago
0
Add comments to _PipelineStage

#1111 kwen2501 closed 4 months ago
0
Make PipelineStage private

#1110 kwen2501 closed 4 months ago
0
refactor manual stage, include docs and example

#1109 H-Huang closed 4 months ago
0
PP Tracer doesn't work with fused_rmsnorm

#1108 wconstab opened 5 months ago
2
Infinite recursion on torch.export for PP tracing

#1107 wconstab opened 5 months ago
0
FSDP+PP requires changing layer iteration code

#1106 wconstab closed 5 months ago
1
FSDP+PP bug where reshard_after_forward must be true

#1105 wconstab opened 5 months ago
6
FSDP+PP tracer issue with cast-to-bf16

#1104 wconstab opened 5 months ago
9
Torchtitan Pipeline Parallel Issue Tracker

#1103 wconstab opened 5 months ago
0
Fix auto-split

#1102 kwen2501 closed 5 months ago
0
Use non-strict mode by default

#1101 kwen2501 opened 5 months ago
1
Add tests for input check

#1100 H-Huang closed 5 months ago
0
Make IR private

#1099 kwen2501 closed 5 months ago
1
Fix interleaved 1f1b race

#1098 H-Huang closed 5 months ago
0
Follow FSDP name change

#1097 kwen2501 closed 5 months ago
0
Make Reducer class private

#1096 kwen2501 closed 5 months ago
0
pipeline() API accepting split_spec

#1095 kwen2501 closed 5 months ago
0
Retrieving the Trained Model

#1094 dheerj188 opened 5 months ago
6
Make some modules private

#1093 kwen2501 closed 5 months ago
0