tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-18 00:27:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

context : add save/load for recurrent context

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-02-21 10:28:42 +02:00
parent 08011c2ca1
commit 2645a7d9a9
2 changed files with 44 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/llama-context.h
View File

@ -525,6 +525,12 @@ public:
bool worst_case) override;
protected:
virtual size_t state_get_data(llama_io_write_i & io) override;
virtual size_t state_set_data(llama_io_read_i & io) override;
virtual size_t state_seq_get_data(llama_io_write_i & io, llama_seq_id seq_id) override;
virtual size_t state_seq_set_data(llama_io_read_i & io, llama_seq_id seq_id) override;
virtual void input_set(const llama_ubatch & ubatch) override;
// TODO: change name to something more meaningful -- does "KV cache" make sense for recurrent models?