Commit d9e61ad
Create a forked KV IO transformer for exporting coreML Llama (pytorch#7778)
Summary:
Create a forked KV IO transformer for exporting coreML Llama
- As discussed in the group chat "
Design: KV cache IO on ANE: llama_transformer vs. static_llama", we agreed that forking a KV IO version llama_transformer is best for code quality and coreML/ANE development purposes
Differential Revision: D684240121 parent 948fba6 commit d9e61ad
File tree
3 files changed
+601
-0
lines changed- examples/models
- kv_io_llama
- llama
3 files changed
+601
-0
lines changed
0 commit comments