Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[shardformer] added embedding gradient check #4124

Merged
merged 1 commit into from
Jun 30, 2023
Merged

[shardformer] added embedding gradient check #4124

merged 1 commit into from
Jun 30, 2023

Conversation

FrankLeeeee
Copy link
Contributor

📌 Checklist before creating the PR

  • I have created an issue for this PR for traceability
  • The title follows the standard format: [doc/gemini/tensor/...]: A concise description
  • I have added relevant tags if possible for us to better distinguish different PRs

🚨 Issue number

Link this PR to your issue with words like fixed to automatically close the linked issue upon merge

e.g. fixed #1234, closed #1234, resolved #1234

Fixed #4123

📝 What does this PR do?

Summarize your work here.
if you have any plots/diagrams/screenshots/tables, please attach them here.

This PR added more testing for embedding gradient and fixed some policies which caused incorrect numerical results.

💥 Checklist before requesting a review

  • I have linked my PR to an issue (instruction)
  • My issue clearly describes the problem/feature/proposal, with diagrams/charts/table/code if possible
  • I have performed a self-review of my code
  • I have added thorough tests.
  • I have added docstrings for all the functions/methods I implemented

⭐️ Do you enjoy contributing to Colossal-AI?

  • 🌝 Yes, I do.
  • 🌚 No, I don't.

Tell us more if you don't enjoy contributing to Colossal-AI.

@FrankLeeeee FrankLeeeee linked an issue Jun 30, 2023 that may be closed by this pull request
[shardformer] add embedding gradient check #4123
Closed
@github-actions
Copy link
Contributor

The code coverage for the changed files is 84%.

Click me to view the complete report 
Name                                                                                       Stmts   Miss  Cover
--------------------------------------------------------------------------------------------------------------
colossalai/auto_parallel/tensor_shard/node_handler/node_handler.py                           164     82    50%
colossalai/auto_parallel/tensor_shard/node_handler/strategy/matmul_strategy_generator.py     388     76    80%
colossalai/auto_parallel/tensor_shard/utils/misc.py                                           45      9    80%
colossalai/checkpoint_io/utils.py                                                            243     43    82%
colossalai/device/device_mesh.py                                                             178     14    92%
colossalai/lazy/lazy_init.py                                                                 299     40    87%
colossalai/nn/layer/base_layer.py                                                             36     15    58%
colossalai/nn/layer/parallel_1d/_operation.py                                                 53     26    51%
colossalai/shardformer/__init__.py                                                             1      0   100%
colossalai/shardformer/_utils.py                                                              42     15    64%
colossalai/shardformer/layer/__init__.py                                                       7      0   100%
colossalai/shardformer/layer/_operation.py                                                   152     51    66%
colossalai/shardformer/layer/dropout.py                                                       35      0   100%
colossalai/shardformer/layer/embedding.py                                                    118      5    96%
colossalai/shardformer/layer/linear.py                                                       156     23    85%
colossalai/shardformer/layer/loss.py                                                          49      8    84%
colossalai/shardformer/layer/normalization.py                                                 50     13    74%
colossalai/shardformer/layer/parallel_module.py                                               76     21    72%
colossalai/shardformer/layer/qkv_fused_linear.py                                             196     29    85%
colossalai/shardformer/layer/utils.py                                                         81     10    88%
colossalai/shardformer/model/__init__.py                                                       0      0   100%
colossalai/shardformer/model/modeling_bert.py                                                 20     20     0%
colossalai/shardformer/policies/__init__.py                                                    0      0   100%
colossalai/shardformer/policies/autopolicy.py                                                 27      2    93%
colossalai/shardformer/policies/basepolicy.py                                                 46      6    87%
colossalai/shardformer/policies/bert.py                                                      117      2    98%
colossalai/shardformer/policies/bloom.py                                                      85      8    91%
colossalai/shardformer/policies/gpt2.py                                                       66      3    95%
colossalai/shardformer/policies/llama.py                                                      37      3    92%
colossalai/shardformer/policies/opt.py                                                        48      2    96%
colossalai/shardformer/policies/t5.py                                                         64      2    97%
colossalai/shardformer/policies/vit.py                                                        23     23     0%
colossalai/shardformer/shard/__init__.py                                                       4      0   100%
colossalai/shardformer/shard/shard_config.py                                                  19      2    89%
colossalai/shardformer/shard/sharder.py                                                       73      3    96%
colossalai/shardformer/shard/shardformer.py                                                   13      0   100%
colossalai/tensor/comm_spec.py                                                               253     93    63%
colossalai/tensor/d_tensor/__init__.py                                                         4      0   100%
colossalai/tensor/d_tensor/api.py                                                            136     18    87%
colossalai/tensor/d_tensor/comm_spec.py                                                      151     35    77%
colossalai/tensor/d_tensor/layout.py                                                          38      1    97%
colossalai/tensor/d_tensor/layout_converter.py                                               195     12    94%
colossalai/tensor/d_tensor/utils.py                                                           38      7    82%
colossalai/tensor/shape_consistency.py                                                       294    120    59%
colossalai/tensor/sharding_spec.py                                                           139     13    91%
colossalai/testing/__init__.py                                                                 4      0   100%
colossalai/testing/comparison.py                                                              54      9    83%
tests/kit/model_zoo/registry.py                                                               18      0   100%
tests/kit/model_zoo/transformers/__init__.py                                                   7      0   100%
tests/kit/model_zoo/transformers/bert.py                                                      42      0   100%
tests/kit/model_zoo/transformers/bloom.py                                                     34      0   100%
tests/kit/model_zoo/transformers/gpt.py                                                       28      0   100%
tests/kit/model_zoo/transformers/llama.py                                                     26      2    92%
tests/kit/model_zoo/transformers/opt.py                                                       32      0   100%
tests/kit/model_zoo/transformers/t5.py                                                        24      0   100%
tests/test_autochunk/test_autochunk_diffuser/test_autochunk_unet.py                           36     10    72%
tests/test_booster/test_mixed_precision/test_fp16_torch.py                                    30      1    97%
tests/test_booster/test_plugin/test_gemini_plugin.py                                          74     10    86%
tests/test_booster/test_plugin/test_low_level_zero_plugin.py                                  60      6    90%
tests/test_booster/test_plugin/test_torch_ddp_plugin.py                                       78      0   100%
tests/test_booster/test_plugin/test_torch_fsdp_plugin.py                                      43      0   100%
tests/test_checkpoint_io/test_gemini_checkpoint_io.py                                         80      0   100%
tests/test_device/test_device_mesh.py                                                         58     36    38%
tests/test_device/test_init_logical_pg.py                                                     27      1    96%
tests/test_fx/test_tracer/test_hf_model/hf_tracer_utils.py                                    21      2    90%
tests/test_fx/test_tracer/test_hf_model/test_hf_albert.py                                     17      1    94%
tests/test_fx/test_tracer/test_hf_model/test_hf_bert.py                                       15      1    93%
tests/test_fx/test_tracer/test_hf_model/test_hf_diffuser.py                                   50     28    44%
tests/test_fx/test_tracer/test_hf_model/test_hf_gpt.py                                        17      1    94%
tests/test_fx/test_tracer/test_hf_model/test_hf_opt.py                                        15      1    93%
tests/test_fx/test_tracer/test_hf_model/test_hf_t5.py                                         17      1    94%
tests/test_fx/test_tracer/test_timm_model/test_timm_model.py                                  36     24    33%
tests/test_fx/test_tracer/test_torchaudio_model/test_torchaudio_model.py                      14      5    64%
tests/test_fx/test_tracer/test_torchrec_model/test_deepfm_model.py                            39      3    92%
tests/test_fx/test_tracer/test_torchrec_model/test_dlrm_model.py                              41      4    90%
tests/test_fx/test_tracer/test_torchvision_model/test_torchvision_model.py                    31      1    97%
tests/test_lazy/lazy_init_utils.py                                                            72     14    81%
tests/test_lazy/test_distribute.py                                                            73      3    96%
tests/test_lazy/test_models.py                                                                13      1    92%
tests/test_shardformer/__init__.py                                                             0      0   100%
tests/test_shardformer/test_layer/test_dist_crossentropy.py                                   27      1    96%
tests/test_shardformer/test_layer/test_dropout.py                                             42      1    98%
tests/test_shardformer/test_layer/test_embedding.py                                           30      1    97%
tests/test_shardformer/test_layer/test_layernorm.py                                           27      1    96%
tests/test_shardformer/test_layer/test_linear_1d.py                                           85      1    99%
tests/test_shardformer/test_layer/test_qkv_fused_linear_1d.py                                 73      1    99%
tests/test_shardformer/test_layer/test_vocab_parallel_embedding_1d.py                         32      1    97%
tests/test_shardformer/test_model/__init__.py                                                  0      0   100%
tests/test_shardformer/test_model/_utils.py                                                   21      0   100%
tests/test_shardformer/test_model/test_shard_bert.py                                          45      1    98%
tests/test_shardformer/test_model/test_shard_bloom.py                                         45      1    98%
tests/test_shardformer/test_model/test_shard_gpt2.py                                          45      1    98%
tests/test_shardformer/test_model/test_shard_llama.py                                         47      1    98%
tests/test_shardformer/test_model/test_shard_opt.py                                           48      1    98%
tests/test_shardformer/test_model/test_shard_t5.py                                            50      1    98%
tests/test_shardformer/test_model/test_shard_vit.py                                           35     20    43%
tests/test_shardformer/test_with_torch_ddp.py                                                 46      2    96%
tests/test_tensor/test_dtensor/test_comm_spec.py                                              78      1    99%
tests/test_tensor/test_dtensor/test_dtensor.py                                                65      5    92%
tests/test_tensor/test_dtensor/test_layout_converter.py                                       91      1    99%
tests/test_tensor/test_shape_consistency.py                                                   50      2    96%
tests/test_tensor/test_sharded_linear.py                                                     130      1    99%
tests/test_tensor/test_sharding_spec.py                                                       13      1    92%
--------------------------------------------------------------------------------------------------------------
TOTAL                                                                                       6610   1066    84%

@flybird11111 flybird11111 merged commit 26ecfd7 into hpcaitech:feature/shardformer Jun 30, 2023
6 checks passed
@FrankLeeeee FrankLeeeee deleted the feature/embedding-grad-check branch June 30, 2023 08:19
@FrankLeeeee FrankLeeeee mentioned this pull request Jun 30, 2023
Closed
flybird11111 pushed a commit to flybird11111/ColossalAI that referenced this pull request Jul 3, 2023
@FrankLeeeee @flybird11111
[shardformer] added embedding gradient check (hpcaitech#4124)
5a42a7d
FrankLeeeee added a commit that referenced this pull request Jul 4, 2023
@FrankLeeeee
[shardformer] added embedding gradient check (#4124)
ae035d3
ver217 pushed a commit to ver217/ColossalAI that referenced this pull request Jul 13, 2023
@FrankLeeeee @ver217
[shardformer] added embedding gradient check (hpcaitech#4124)
a1653e6
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@flybird11111 flybird11111 flybird11111 approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

[shardformer] add embedding gradient check
2 participants
@FrankLeeeee @flybird11111