llama.cpp: Embedding mit -ub 4096 (ub 8192 braucht 5 GB Rechenpuffer auf der P2000)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-19 17:44:15 +02:00
parent 986f1116fe
commit 862825fa1e
3 changed files with 4 additions and 3 deletions
@@ -36,4 +36,3 @@
  loop: "{{ llamacpp_profiles | map('extract', llamacpp_models) | flatten }}"
  loop_control:
    label: "{{ item.dest }}"
-