tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-07-18 08:37:43 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : introduce llama_io interfaces

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-02-13 12:18:44 +02:00
parent fbe6a07256
commit 3a504d9a0b
7 changed files with 250 additions and 334 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
src/llama-context.h
View File

@ -15,6 +15,9 @@
#include <vector>
#include <set>
class llama_io_read_i;
class llama_io_write_i;
using llama_loras = std::unordered_map<struct llama_adapter_lora *, float>;
struct llama_context : public llama_graph_i {
@ -178,9 +181,10 @@ struct llama_context : public llama_graph_i {
virtual llama_perf_context_data perf_get_data() const;
virtual void perf_reset();
protected:
// members
protected:
const llama_model & model;
llama_cparams cparams;
@ -502,11 +506,11 @@ public:
size_t n_token_count) override;
private:
size_t state_get_data(struct llama_data_write & data_ctx);
size_t state_set_data(struct llama_data_read & data_ctx);
size_t state_get_data(llama_io_write_i & io);
size_t state_set_data(llama_io_read_i & io);
size_t state_seq_get_data(struct llama_data_write & data_ctx, llama_seq_id seq_id);
size_t state_seq_set_data(struct llama_data_read & data_ctx, llama_seq_id seq_id);
size_t state_seq_get_data(llama_io_write_i & io, llama_seq_id seq_id);
size_t state_seq_set_data(llama_io_read_i & io, llama_seq_id seq_id);
};
// For internal test use