Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-01-01 00:39:00 +01:00
Code Issues Packages Projects Releases Wiki Activity

llama : fix comments for llama_kv_cache API

Browse Source
This commit is contained in:
Georgi Gerganov 2023-10-03 21:01:45 +03:00
parent 337120cc0d
commit 5418932b71
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
llama.h
View File

@ -330,16 +330,16 @@ extern "C" {
"avoid using this, it will be removed in the future, instead - count the tokens in user code"); "avoid using this, it will be removed in the future, instead - count the tokens in user code");
// Remove all tokens data of cells in [c0, c1) // Remove all tokens data of cells in [c0, c1)
// c0 < -1 : [0, c1] // c0 < 0 : [0, c1]
// c1 < -1 : [c0, inf) // c1 < 0 : [c0, inf)
LLAMA_API void llama_kv_cache_tokens_rm( LLAMA_API void llama_kv_cache_tokens_rm(
struct llama_context * ctx, struct llama_context * ctx,
int32_t c0, int32_t c0,
int32_t c1); int32_t c1);
// Removes all tokens that belong to the specified sequence and have positions in [p0, p1) // Removes all tokens that belong to the specified sequence and have positions in [p0, p1)
// p0 < -1 : [0, p1] // p0 < 0 : [0, p1]
// p1 < -1 : [p0, inf) // p1 < 0 : [p0, inf)
LLAMA_API void llama_kv_cache_seq_rm( LLAMA_API void llama_kv_cache_seq_rm(
struct llama_context * ctx, struct llama_context * ctx,
llama_seq_id seq_id, llama_seq_id seq_id,
@ -348,8 +348,8 @@ extern "C" {
// Copy all tokens that belong to the specified sequence to another sequence // Copy all tokens that belong to the specified sequence to another sequence
// Note that this does not allocate extra KV cache memory - it simply assigns the tokens to the new sequence // Note that this does not allocate extra KV cache memory - it simply assigns the tokens to the new sequence
// p0 < -1 : [0, p1] // p0 < 0 : [0, p1]
// p1 < -1 : [p0, inf) // p1 < 0 : [p0, inf)
LLAMA_API void llama_kv_cache_seq_cp( LLAMA_API void llama_kv_cache_seq_cp(
struct llama_context * ctx, struct llama_context * ctx,
llama_seq_id seq_id_src, llama_seq_id seq_id_src,
@ -364,8 +364,8 @@ extern "C" {
// Adds relative position "delta" to all tokens that belong to the specified sequence and have positions in [p0, p1) // Adds relative position "delta" to all tokens that belong to the specified sequence and have positions in [p0, p1)
// If the KV cache is RoPEd, the KV data is updated accordingly // If the KV cache is RoPEd, the KV data is updated accordingly
// p0 < -1 : [0, p1] // p0 < 0 : [0, p1]
// p1 < -1 : [p0, inf) // p1 < 0 : [p0, inf)
LLAMA_API void llama_kv_cache_seq_shift( LLAMA_API void llama_kv_cache_seq_shift(
struct llama_context * ctx, struct llama_context * ctx,
llama_seq_id seq_id, llama_seq_id seq_id,