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

fix whitespace

Browse source
This commit is contained in:
Pavol Rusnak 2023-04-02 12:18:54 +02:00
parent a463fb7668
commit 4912f9d5f0
No known key found for this signature in database
GPG key ID: 91F3B339B9A02A3D
2 changed files with 13 additions and 13 deletions
Download patch file Download diff file Expand all files Collapse all files

20
llama.cpp
View file

@ -1671,28 +1671,28 @@ int llama_model_quantize(
// Returns the KV cache that will contain the context for the
// ongoing prediction with the model.
const uint8_t * llama_get_kv_cache(struct llama_context * ctx) {
return ctx->model.kv_self.buf.data();
return ctx->model.kv_self.buf.data();
}
// Returns the size of the KV cache
size_t llama_get_kv_cache_size(struct llama_context * ctx) {
return ctx->model.kv_self.buf.size();
return ctx->model.kv_self.buf.size();
}
int llama_get_kv_cache_token_count(struct llama_context * ctx) {
return ctx->model.kv_self.n;
return ctx->model.kv_self.n;
}
// Sets the KV cache containing the current context for the model
void llama_set_kv_cache(
struct llama_context * ctx,
const uint8_t * kv_cache,
size_t n_size,
int n_token_count) {
// Make sure we have the same kv cache setup
LLAMA_ASSERT(ctx->model.kv_self.buf.size() == n_size);
memcpy(ctx->model.kv_self.buf.data(), kv_cache, n_size);
ctx->model.kv_self.n = n_token_count;
const uint8_t * kv_cache,
size_t n_size,
int n_token_count) {
// Make sure we have the same kv cache setup
LLAMA_ASSERT(ctx->model.kv_self.buf.size() == n_size);
memcpy(ctx->model.kv_self.buf.data(), kv_cache, n_size);
ctx->model.kv_self.n = n_token_count;
}
int llama_eval(

6
llama.h
View file

@ -96,9 +96,9 @@ extern "C" {
// Sets the KV cache containing the current context for the model
LLAMA_API void llama_set_kv_cache(
struct llama_context * ctx,
const uint8_t * kv_cache,
size_t n_size,
int n_token_count);
const uint8_t * kv_cache,
size_t n_size,
int n_token_count);
// Run the llama inference to obtain the logits and probabilities for the next token.
// tokens + n_tokens is the provided batch of new tokens to process