Public/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-19 17:24:57 +02:00
Code Issues Packages Projects Releases Wiki Activity

rm debugging

Browse Source
This commit is contained in:
Xuan Son Nguyen
2026-09-01 03:06:19 +02:00
parent 4b51a5f2aa
commit d692634fa7
1 changed files with 0 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/mtmd/clip.cpp
-16
View File
@@ -5874,22 +5874,6 @@ bool clip_encode(struct clip_ctx * ctx, struct clip_encode_params * params) {
LOG_INF("\n=== MTMD_DEBUG_EMBEDDINGS ===\n");
LOG_INF("Shape: [%lld, %lld]\n", (long long)n_embd, (long long)n_tokens);
// when the env var holds a path (any value other than "1"), also dump the raw data
// format: [int32 n_tokens][int32 n_embd][f32 data (token-major)]
const char * dump_path = std::getenv("MTMD_DEBUG_EMBEDDINGS");
if (dump_path && strcmp(dump_path, "1") != 0) {
FILE * f = fopen(dump_path, "wb");
if (f) {
const int32_t hdr[2] = { (int32_t)n_tokens, (int32_t)n_embd };
fwrite(hdr, sizeof(hdr), 1, f);
fwrite(emb_data.data(), sizeof(float), emb_data.size(), f);
fclose(f);
LOG_INF("Dumped raw embeddings to %s\n", dump_path);
} else {
LOG_ERR("Failed to open %s for writing\n", dump_path);
}
}
// Print first few values of first token
LOG_INF("Token 0 (first 16 values): ");
for (int i = 0; i < std::min((int64_t)16, n_embd); i++) {