tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-14 20:29:41 -04:00
Code Issues Packages Projects Releases Wiki Activity

embeddings: fix extraction of CLS pooling results (#14927)

Browse Source 
* embeddings: fix extraction of CLS pooling results

* merge RANK pooling into CLS case for inputs
This commit is contained in:
Douglas Hanley
2025-07-30 00:25:05 -05:00
committed by GitHub
parent 61550f8231
commit a118d80233
1 changed files with 16 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
src/llama-graph.cpp
View File

@@ -188,12 +188,12 @@ void llm_graph_input_mean::set_input(const llama_ubatch * ubatch) {
void llm_graph_input_cls::set_input(const llama_ubatch * ubatch) { void llm_graph_input_cls::set_input(const llama_ubatch * ubatch) {
const int64_t n_tokens = ubatch->n_tokens; const int64_t n_tokens = ubatch->n_tokens;
const int64_t n_seq_tokens = ubatch->n_seq_tokens;
const int64_t n_seqs_unq = ubatch->n_seqs_unq; const int64_t n_seqs_unq = ubatch->n_seqs_unq;
if (cparams.embeddings && ( if (cparams.embeddings && (
cparams.pooling_type == LLAMA_POOLING_TYPE_CLS || cparams.pooling_type == LLAMA_POOLING_TYPE_CLS ||
cparams.pooling_type == LLAMA_POOLING_TYPE_RANK cparams.pooling_type == LLAMA_POOLING_TYPE_RANK ||
cparams.pooling_type == LLAMA_POOLING_TYPE_LAST
)) { )) {
GGML_ASSERT(cls); GGML_ASSERT(cls);
GGML_ASSERT(ggml_backend_buffer_is_host(cls->buffer)); GGML_ASSERT(ggml_backend_buffer_is_host(cls->buffer));
@@ -201,25 +201,10 @@ void llm_graph_input_cls::set_input(const llama_ubatch * ubatch) {
uint32_t * data = (uint32_t *) cls->data; uint32_t * data = (uint32_t *) cls->data;
memset(cls->data, 0, n_seqs_unq*ggml_element_size(cls)); memset(cls->data, 0, n_seqs_unq*ggml_element_size(cls));
for (int i = 0; i < n_tokens; i += n_seq_tokens) { std::vector<int> target_pos(n_seqs_unq, -1);
for (int s = 0; s < ubatch->n_seq_id[i]; ++s) { std::vector<int> target_row(n_seqs_unq, -1);
const llama_seq_id seq_id = ubatch->seq_id[i][s];
const int32_t seq_idx = ubatch->seq_idx[seq_id];
data[seq_idx] = i; bool last = cparams.pooling_type == LLAMA_POOLING_TYPE_LAST;
}
}
}
if (cparams.embeddings && cparams.pooling_type == LLAMA_POOLING_TYPE_LAST) {
GGML_ASSERT(cls);
GGML_ASSERT(ggml_backend_buffer_is_host(cls->buffer));
uint32_t * data = (uint32_t *) cls->data;
memset(cls->data, 0, n_seqs_unq*ggml_element_size(cls));
std::vector<int> last_pos(n_seqs_unq, -1);
std::vector<int> last_row(n_seqs_unq, -1);
for (int i = 0; i < n_tokens; ++i) { for (int i = 0; i < n_tokens; ++i) {
const llama_pos pos = ubatch->pos[i]; const llama_pos pos = ubatch->pos[i];
@@ -228,16 +213,20 @@ void llm_graph_input_cls::set_input(const llama_ubatch * ubatch) {
const llama_seq_id seq_id = ubatch->seq_id[i][s]; const llama_seq_id seq_id = ubatch->seq_id[i][s];
const int32_t seq_idx = ubatch->seq_idx[seq_id]; const int32_t seq_idx = ubatch->seq_idx[seq_id];
if (pos >= last_pos[seq_idx]) { if (
last_pos[seq_idx] = pos; (target_pos[seq_idx] == -1) ||
last_row[seq_idx] = i; ( last && pos >= target_pos[seq_idx]) ||
(!last && pos < target_pos[seq_idx])
) {
target_pos[seq_idx] = pos;
target_row[seq_idx] = i;
} }
} }
} }
for (int s = 0; s < n_seqs_unq; ++s) { for (int s = 0; s < n_seqs_unq; ++s) {
if (last_row[s] >= 0) { if (target_row[s] >= 0) {
data[s] = last_row[s]; data[s] = target_row[s];
} }
} }
} }