ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

context : fix state io for memory-less contexts (#13470)

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2025-05-12 15:12:27 +03:00 • committed by GitHub
parent 91159ee9df
commit 064cc596ac
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 14 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

21
src/llama-context.cpp
View file

@ -1788,10 +1788,13 @@ size_t llama_context::state_read_data(llama_io_read_i & io) {
}
}
LLAMA_LOG_DEBUG("%s: - reading KV self\n", __func__);
llama_kv_cache * kv_self = static_cast<llama_kv_cache *>(memory.get());
if (memory) {
LLAMA_LOG_DEBUG("%s: - reading KV self\n", __func__);
kv_self->state_read(io);
llama_kv_cache * kv_self = static_cast<llama_kv_cache *>(memory.get());
kv_self->state_read(io);
}
return io.n_bytes();
}
@ -1799,9 +1802,11 @@ size_t llama_context::state_read_data(llama_io_read_i & io) {
size_t llama_context::state_seq_write_data(llama_io_write_i & io, llama_seq_id seq_id) {
GGML_UNUSED(seq_id);
llama_kv_cache * kv_self = static_cast<llama_kv_cache *>(memory.get());
if (memory) {
llama_kv_cache * kv_self = static_cast<llama_kv_cache *>(memory.get());
kv_self->state_write(io, seq_id);
kv_self->state_write(io, seq_id);
}
return io.n_bytes();
}
@ -1809,9 +1814,11 @@ size_t llama_context::state_seq_write_data(llama_io_write_i & io, llama_seq_id s
size_t llama_context::state_seq_read_data(llama_io_read_i & io, llama_seq_id seq_id) {
GGML_UNUSED(seq_id);
llama_kv_cache * kv_self = static_cast<llama_kv_cache *>(memory.get());
if (memory) {
llama_kv_cache * kv_self = static_cast<llama_kv_cache *>(memory.get());
kv_self->state_read(io, seq_id);
kv_self->state_read(io, seq_id);
}
return io.n_bytes();
}