search

cli99 / llm-analysis

Latency and Memory Analysis of Transformer Models for Training and Inference
Apache License 2.0
343 stars 40 forks source link

fix memory calculation for optimizer states, gradients, and activation #11

Closed cli99 closed 11 months ago

codecov-commenter commented 11 months ago

Codecov Report

Attention: 3 lines in your changes are missing coverage. Please review.

Comparison is base (4df2a89) 81.63% compared to head (934ef7d) 81.78%.

:exclamation: Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files ```diff @@ Coverage Diff @@ ## main #11 +/- ## ========================================== + Coverage 81.63% 81.78% +0.15% ========================================== Files 8 8 Lines 969 977 +8 ========================================== + Hits 791 799 +8 Misses 178 178 ``` | [Files](https://app.codecov.io/gh/cli99/llm-analysis/pull/11?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=Cheng+Li) | Coverage Δ | | |---|---|---| | [llm\_analysis/analysis.py](https://app.codecov.io/gh/cli99/llm-analysis/pull/11?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=Cheng+Li#diff-bGxtX2FuYWx5c2lzL2FuYWx5c2lzLnB5) | `78.22% <ø> (+0.21%)` | :arrow_up: | | [llm\_analysis/constant.py](https://app.codecov.io/gh/cli99/llm-analysis/pull/11?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=Cheng+Li#diff-bGxtX2FuYWx5c2lzL2NvbnN0YW50LnB5) | `100.00% <100.00%> (ø)` | | | [llm\_analysis/utils.py](https://app.codecov.io/gh/cli99/llm-analysis/pull/11?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=Cheng+Li#diff-bGxtX2FuYWx5c2lzL3V0aWxzLnB5) | `90.90% <100.00%> (ø)` | | | [tests/test\_inference.py](https://app.codecov.io/gh/cli99/llm-analysis/pull/11?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=Cheng+Li#diff-dGVzdHMvdGVzdF9pbmZlcmVuY2UucHk=) | `100.00% <100.00%> (ø)` | | | [tests/test\_training.py](https://app.codecov.io/gh/cli99/llm-analysis/pull/11?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=Cheng+Li#diff-dGVzdHMvdGVzdF90cmFpbmluZy5weQ==) | `100.00% <100.00%> (ø)` | | | [llm\_analysis/config.py](https://app.codecov.io/gh/cli99/llm-analysis/pull/11?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=Cheng+Li#diff-bGxtX2FuYWx5c2lzL2NvbmZpZy5weQ==) | `67.89% <76.92%> (+0.16%)` | :arrow_up: |

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.