PR #18181 Reduce LLaMA memory usage

Reduce LLaMA memory usage #18181

kunal-vaishnavi merged 7 commits into microsoft:main from kunal-vaishnavi:kvaishnavi/llama-reduce-memory-usage

Reduce memory usage during export and benchmarking

53a6cef6

github-advanced-security commented on 2023-10-31

Add changes suggested by linter

3c02b99b

kunal-vaishnavi added release:1.16.2

kunal-vaishnavi added sdxl_llama

Fix CodeQL error

094e58ca

frank-dong-ms commented on 2023-10-31

Update max sequence length

d998c3cf

Add changes from PR feedback

428280e7

frank-dong-ms commented on 2023-10-31

frank-dong-ms dismissed these changes on 2023-10-31

Update max sequence lengths

019556e1

kunal-vaishnavi dismissed their stale review via 019556e1 2 years ago

Remove max sequence length as optional argument

f2c61d88

tianleiwu approved these changes on 2023-11-01

kunal-vaishnavi merged d1b85f5f into main 2 years ago

tianleiwu removed release:1.16.2

tianleiwu removed sdxl_llama

Reviewers

tianleiwu

frank-dong-ms

Assignees

No one assigned

Labels

None yet

Milestone

No milestone