| 370 | /* ── Dense vector operations ─────────────────────────────────────── */ |
| 371 | |
| 372 | float cbm_sem_cosine(const cbm_sem_vec_t *a, const cbm_sem_vec_t *b) { |
| 373 | if (!a || !b) { |
| 374 | return 0.0F; |
| 375 | } |
| 376 | float dot = 0.0F; |
| 377 | float mag_a = 0.0F; |
| 378 | float mag_b = 0.0F; |
| 379 | for (int i = 0; i < CBM_SEM_DIM; i++) { |
| 380 | dot += a->v[i] * b->v[i]; |
| 381 | mag_a += a->v[i] * a->v[i]; |
| 382 | mag_b += b->v[i] * b->v[i]; |
| 383 | } |
| 384 | float denom = sqrtf(mag_a) * sqrtf(mag_b); |
| 385 | if (denom < CBM_SEM_DENOM_EPS) { |
| 386 | return 0.0F; |
| 387 | } |
| 388 | return dot / denom; |
| 389 | } |
| 390 | |
| 391 | /* Pretrained token lookup table — built lazily on first use. */ |
| 392 | static CBMHashTable *g_pretrained_map = NULL; |