extern/whisper.cpp
1
0
Fork 1
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-08-09 19:08:32 +02:00
Code Issues Packages Projects Releases Wiki Activity

talk-llama : sync llama.cpp

Browse Source
This commit is contained in:
Georgi Gerganov
2025-07-01 12:21:09 +03:00
parent c4ea72be9a
commit 1f816de7da
24 changed files with 1456 additions and 430 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
examples/talk-llama/llama-memory-recurrent.h
View File

@ -11,8 +11,8 @@
// llama_memory_recurrent
//
// TODO: extract the cache state used for graph computation into llama_memory_recurrent_state_i
// see the implementation of llama_kv_cache_unified_state_i for an example how to do it
// TODO: extract the cache state used for graph computation into llama_memory_recurrent_context_i
// see the implementation of llama_kv_cache_unified_context_i for an example how to do it
class llama_memory_recurrent : public llama_memory_i {
public:
@ -34,14 +34,14 @@ public:
// llama_memory_i
//
llama_memory_state_ptr init_batch(
llama_memory_context_ptr init_batch(
llama_batch_allocr & balloc,
uint32_t n_ubatch,
bool embd_all) override;
llama_memory_state_ptr init_full() override;
llama_memory_context_ptr init_full() override;
llama_memory_state_ptr init_update(llama_context * lctx, bool optimize) override;
llama_memory_context_ptr init_update(llama_context * lctx, bool optimize) override;
void clear(bool data) override;
@ -125,24 +125,24 @@ private:
bool state_read_data(llama_io_read_i & io, uint32_t cell_count);
};
class llama_memory_recurrent_state : public llama_memory_state_i {
class llama_memory_recurrent_context : public llama_memory_context_i {
public:
// used for errors
llama_memory_recurrent_state(llama_memory_status status);
llama_memory_recurrent_context(llama_memory_status status);
// used to create a full-cache state
llama_memory_recurrent_state(
// used to create a full-cache or update context
llama_memory_recurrent_context(
llama_memory_recurrent * mem);
// used to create a state from a batch
llama_memory_recurrent_state(
// used to create a batch processing context from a batch
llama_memory_recurrent_context(
llama_memory_recurrent * mem,
std::vector<llama_ubatch> ubatches);
virtual ~llama_memory_recurrent_state();
virtual ~llama_memory_recurrent_context();
//
// llama_memory_state_i
// llama_memory_context_i
//
bool next() override;
@ -152,7 +152,7 @@ public:
const llama_ubatch & get_ubatch() const override;
//
// llama_memory_recurrent_state specific API
// llama_memory_recurrent_context specific API
//
uint32_t get_n_rs() const;