diff options
| author | Georgi Gerganov <ggerganov@gmail.com> | 2024-02-25 22:12:24 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-02-25 22:12:24 +0200 |
| commit | bf08e00643fd529f748f0a858fd79f3061e3fa18 (patch) | |
| tree | 0043ee582e83a19c8f1ca6d75d1519038f866e1c /examples/main/main.cpp | |
| parent | f7625019c51ca437a5840576d92362cfa710e4a2 (diff) | |
llama : refactor k-shift implementation + KV defragmentation (#5691)
* llama : refactor k-shift implementation
ggml-ci
* llama : rename llama_kv_cache_seq_shift to llama_kv_cache_seq_add
* llama : cont k-shift refactoring + normalize type names
ggml-ci
* minor : fix MPI builds
* llama : reuse n_rot from the build context
ggml-ci
* llama : revert enum name changes from this PR
ggml-ci
* llama : update llama_rope_type
* llama : add comment about rope values
* llama : fix build
* passkey : apply kv cache updates explicitly
ggml-ci
* llama : change name to llama_kv_cache_update()
* llama : add llama_kv_cache_seq_pos_max()
* passkey : fix llama_kv_cache_seq_pos_max() usage
* llama : some llama_kv_cell simplifications
* llama : add llama_kv_cache_compress (EXPERIMENTAL)
* llama : add alternative KV cache merging (EXPERIMENTAL)
* llama : add llama_kv_cache_defrag
* llama : comments
* llama : remove llama_kv_cache_compress
will add in a separate PR
ggml-ci
* llama : defragment via non-overlapping moves
* llama : ggml_graph based defrag implementation
ggml-ci
* llama : switch the loop order in build_defrag
* llama : add comments
Diffstat (limited to 'examples/main/main.cpp')
| -rw-r--r-- | examples/main/main.cpp | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/examples/main/main.cpp b/examples/main/main.cpp index 7555dffe..34e84d0d 100644 --- a/examples/main/main.cpp +++ b/examples/main/main.cpp @@ -548,8 +548,8 @@ int main(int argc, char ** argv) { LOG("context full, swapping: n_past = %d, n_left = %d, n_ctx = %d, n_keep = %d, n_discard = %d\n", n_past, n_left, n_ctx, params.n_keep, n_discard); - llama_kv_cache_seq_rm (ctx, 0, params.n_keep , params.n_keep + n_discard); - llama_kv_cache_seq_shift(ctx, 0, params.n_keep + n_discard, n_past, -n_discard); + llama_kv_cache_seq_rm (ctx, 0, params.n_keep , params.n_keep + n_discard); + llama_kv_cache_seq_add(ctx, 0, params.n_keep + n_discard, n_past, -n_discard); n_past -= n_discard; @@ -576,9 +576,9 @@ int main(int argc, char ** argv) { LOG("div: [%6d, %6d] / %6d -> [%6d, %6d]\n", ga_i + ib*bd, ga_i + ib*bd + ga_w, ga_n, (ga_i + ib*bd)/ga_n, (ga_i + ib*bd + ga_w)/ga_n); LOG("shift: [%6d, %6d] + %6d -> [%6d, %6d]\n", ga_i + ib*bd + ga_w, n_past + ib*bd, dd, ga_i + ib*bd + ga_w + dd, n_past + ib*bd + dd); - llama_kv_cache_seq_shift(ctx, 0, ga_i, n_past, ib*bd); - llama_kv_cache_seq_div (ctx, 0, ga_i + ib*bd, ga_i + ib*bd + ga_w, ga_n); - llama_kv_cache_seq_shift(ctx, 0, ga_i + ib*bd + ga_w, n_past + ib*bd, dd); + llama_kv_cache_seq_add(ctx, 0, ga_i, n_past, ib*bd); + llama_kv_cache_seq_div(ctx, 0, ga_i + ib*bd, ga_i + ib*bd + ga_w, ga_n); + llama_kv_cache_seq_add(ctx, 0, ga_i + ib*bd + ga_w, n_past + ib*bd, dd); n_past -= bd; |