Michael and Claude Opus 5.5
539e36ea49
Spéculation : SPEC=ngram et mtp+ngram, n-grammes du contexte en opt-in
...
En mode Code, le modèle recopie sans cesse ce qui est déjà dans le
contexte : chemins, diffs, arguments JSON, fichiers réécrits. ngram-mod
de llama.cpp y cherche la suite des 24 derniers jetons et propose d'un
coup les 48 à 64 suivants, sans modèle brouillon. La clé SPEC (off par
défaut, inchangé) gagne deux valeurs : ngram, et mtp+ngram qui y ajoute
la tête MTP. Même vérification exacte que MTP : distribution inchangée,
mais pas le même texte au bit près (vérification par lots), donc on
compare des sessions rejouées, pas des diffs.
- forme explicite seulement : --spec-type ngram-mod
--spec-ngram-mod-n-match 24 --spec-ngram-mod-n-min 48
--spec-ngram-mod-n-max 64 ; jamais --spec-default (TODO amont, contenu
susceptible de changer avec le moteur)
- garde-fou d'aide sur « --spec-ngram-mod-n-match » (pas le simple mot
ngram-mod, encore listé par les moteurs d'avant le renommage) ; absent =
SPEC ignoré, note au journal
- mtp+ngram : --spec-type draft-mtp,ngram-mod seulement avec draft-mtp
dans l'aide ET une tête MTP réelle (tenseur nextn ou MODEL_DRAFT) ;
sinon n-grammes seuls, avec la raison — jamais « failed to create MTP
context » en boucle
- EXTRA_ARGS/LLAMA_ARG_* : mêmes exclusions qu'avant, plus --spec_type et
tout --spec-ngram-* (--spec-type s'additionne sans prévenir) ; pour
ngram seul, un --spec-draft-* fait aussi taire Loki
- poids sur CPU (experts MoE, -ot, NGL partiel) : refusé tant que
GGML_OP_OFFLOAD_MIN_BATCH (env ou OP_OFFLOAD_MIN_BATCH) ne dépasse pas
le lot de vérification de 65 jetons, avec la suggestion
OP_OFFLOAD_MIN_BATCH=128 ; avertissements pour un modèle plus gros que
la RAM, les points de reprise d'un hybride, NGL imposé (logits)
- avec ngram, une tête MTP ou MODEL_DRAFT inutilisés sont signalés
- --spec-synth-len/--spec-synth-rates (et LLAMA_ARG_SPEC_SYNTH_LEN) :
acceptation au hasard, avertissement au lancement ; Loki ne les pose
jamais
- télémétrie : /api/perf/summary donne draft_n, draft_accepted et
draft_rate par nature de requête
- UI : deux options du sélecteur de décodage spéculatif ; README et
configTemplate
- tests : forme explicite, garde d'aide, repli de mtp+ngram, exclusions,
seuil d'offload, ligne inchangée sans SPEC, acceptation par nature
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-10-04 10:28:25 +02:00
..
2026-10-02 23:30:00 +02:00
2026-10-04 10:28:25 +02:00
2026-08-14 21:33:15 +00:00
2026-08-20 09:31:59 +00:00
2026-08-14 21:37:51 +00:00
2026-08-16 11:02:35 +00:00
2026-08-14 21:37:51 +00:00
2026-08-20 09:31:59 +00:00
2026-08-16 20:43:47 +00:00
2026-08-18 21:58:04 +02:00
2026-10-03 23:57:08 +02:00
2026-10-04 02:52:54 +02:00
2026-10-02 23:13:56 +02:00
2026-10-03 23:37:55 +02:00
2026-10-03 23:37:55 +02:00
2026-08-14 21:33:15 +00:00
2026-08-20 10:32:44 +00:00
2026-08-20 10:32:44 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-19 08:28:53 +02:00
2026-08-14 21:33:15 +00:00
2026-08-29 17:53:14 +02:00
2026-10-04 00:16:41 +02:00
2026-10-04 00:16:41 +02:00
2026-09-12 20:20:33 +00:00
2026-08-14 21:33:15 +00:00
2026-09-12 20:20:33 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-16 20:43:47 +00:00
2026-10-03 22:43:04 +02:00
2026-10-04 10:17:08 +02:00
2026-08-20 09:31:59 +00:00
2026-10-03 23:04:08 +02:00
2026-10-04 10:17:08 +02:00
2026-10-04 10:17:08 +02:00
2026-10-03 23:57:08 +02:00
2026-10-04 08:26:04 +02:00
2026-10-03 22:58:23 +02:00
2026-10-04 00:09:24 +02:00
2026-10-03 23:11:31 +02:00
2026-10-04 10:28:25 +02:00
2026-10-03 22:27:04 +02:00
2026-10-04 10:17:08 +02:00
2026-10-03 22:47:38 +02:00
2026-10-03 22:53:44 +02:00
2026-10-04 10:02:40 +02:00
2026-10-04 10:02:40 +02:00
2026-10-04 10:28:25 +02:00
2026-10-04 10:28:25 +02:00
2026-10-03 22:37:59 +02:00
2026-10-03 22:37:59 +02:00
2026-10-04 10:28:25 +02:00
2026-09-22 13:59:10 +00:00
2026-09-22 13:59:10 +00:00
2026-09-22 13:19:37 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 08:26:04 +02:00
2026-10-04 09:25:35 +02:00
2026-10-04 10:02:40 +02:00
2026-10-02 23:23:21 +02:00
2026-10-04 01:04:03 +02:00
2026-10-04 10:02:40 +02:00
2026-09-12 20:25:48 +00:00
2026-10-02 23:07:46 +02:00
2026-10-04 09:46:01 +02:00
2026-08-16 15:33:55 +00:00
2026-09-22 13:59:10 +00:00
2026-10-02 23:04:30 +02:00
2026-10-02 23:04:30 +02:00
2026-08-30 15:47:10 +02:00
2026-10-04 08:44:46 +02:00
2026-10-02 23:11:40 +02:00
2026-10-02 23:10:56 +02:00
2026-10-02 23:10:56 +02:00
2026-10-02 23:08:46 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 09:49:06 +02:00
2026-10-04 09:49:06 +02:00
2026-10-02 23:04:51 +02:00
2026-10-02 23:04:51 +02:00
2026-08-30 15:47:10 +02:00
2026-10-02 23:05:08 +02:00
2026-10-04 02:27:54 +02:00
2026-10-04 02:27:54 +02:00
2026-10-04 09:01:57 +02:00
2026-10-04 10:17:08 +02:00
2026-10-04 09:21:57 +02:00
2026-10-04 09:46:01 +02:00
2026-10-04 01:34:36 +02:00
2026-10-02 23:11:40 +02:00
2026-09-07 22:34:33 +02:00
2026-09-22 13:59:10 +00:00
2026-09-07 22:33:55 +02:00
2026-10-02 23:13:56 +02:00
2026-09-22 13:33:39 +00:00
2026-09-22 13:33:39 +00:00
2026-10-04 09:01:57 +02:00
2026-10-02 23:34:42 +02:00
2026-10-02 23:34:42 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 08:44:46 +02:00
2026-10-04 08:26:04 +02:00
2026-10-04 08:26:04 +02:00
2026-09-12 21:05:19 +00:00
2026-09-22 13:19:37 +00:00
2026-08-16 15:33:55 +00:00
2026-08-20 09:31:59 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:47:57 +00:00
2026-10-02 23:32:29 +02:00
2026-10-02 23:32:29 +02:00
2026-08-18 21:58:57 +02:00
2026-10-02 23:30:00 +02:00
2026-10-02 23:38:07 +02:00
2026-10-02 23:36:51 +02:00
2026-10-02 23:36:51 +02:00
2026-08-20 09:31:59 +00:00
2026-08-18 21:58:57 +02:00
2026-08-20 09:31:59 +00:00
2026-08-18 21:58:57 +02:00
2026-08-20 09:31:59 +00:00
2026-10-02 23:40:38 +02:00
2026-10-02 23:31:33 +02:00
2026-10-04 01:48:49 +02:00
2026-10-04 09:01:57 +02:00
2026-09-22 14:08:01 +00:00
2026-10-04 00:29:17 +02:00
2026-10-02 23:36:01 +02:00
2026-08-20 09:31:59 +00:00
2026-08-18 21:58:57 +02:00
2026-10-02 23:30:00 +02:00
2026-10-02 23:30:00 +02:00
2026-10-04 08:44:46 +02:00
2026-09-22 13:19:37 +00:00
2026-09-22 13:19:37 +00:00
2026-09-22 13:19:37 +00:00
2026-09-22 13:19:37 +00:00
2026-10-04 00:53:38 +02:00
2026-09-07 22:34:57 +02:00
2026-09-07 22:34:57 +02:00
2026-09-07 22:34:57 +02:00
2026-09-07 22:34:57 +02:00
2026-09-07 22:34:57 +02:00
2026-09-07 22:34:57 +02:00
2026-09-07 22:34:57 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 03:09:01 +02:00
2026-10-04 02:52:54 +02:00
2026-10-04 03:09:01 +02:00
2026-10-04 10:02:40 +02:00
2026-08-18 08:56:32 +00:00
2026-10-02 23:36:01 +02:00
2026-10-04 10:17:08 +02:00
2026-08-20 08:49:09 +00:00
2026-08-20 08:49:09 +00:00
2026-10-04 01:25:39 +02:00
2026-10-04 09:46:01 +02:00
2026-08-16 19:45:43 +00:00
2026-10-04 10:02:40 +02:00
2026-10-04 08:26:04 +02:00
2026-10-04 08:26:04 +02:00
2026-08-14 21:33:15 +00:00
2026-10-02 23:18:35 +02:00
2026-10-04 01:16:48 +02:00
2026-10-02 23:18:35 +02:00
2026-10-04 10:02:40 +02:00
2026-10-04 10:02:40 +02:00
2026-10-04 10:17:08 +02:00
2026-10-04 10:17:08 +02:00
2026-10-03 23:44:33 +02:00
2026-10-04 10:02:40 +02:00
2026-08-15 14:36:04 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-18 21:58:57 +02:00
2026-10-02 23:34:42 +02:00
2026-08-18 10:53:09 +00:00
2026-08-17 10:06:14 +02:00
2026-08-17 10:06:14 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 01:34:36 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 01:34:36 +02:00
2026-09-22 13:59:10 +00:00
2026-09-22 13:59:10 +00:00
2026-09-22 13:59:10 +00:00
2026-09-22 13:59:10 +00:00
2026-09-07 22:34:33 +02:00
2026-10-04 08:44:46 +02:00
2026-09-22 13:59:10 +00:00
2026-10-02 23:27:43 +02:00
2026-09-22 13:59:10 +00:00
2026-10-02 23:05:08 +02:00
2026-09-22 13:59:10 +00:00
2026-09-22 13:59:10 +00:00
2026-10-04 02:19:03 +02:00
2026-09-22 13:59:10 +00:00
2026-09-22 13:59:10 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-09-07 22:33:55 +02:00
2026-08-14 21:37:51 +00:00
2026-10-04 01:31:00 +02:00
2026-08-20 09:31:59 +00:00
2026-10-04 10:28:25 +02:00
2026-10-04 10:28:25 +02:00
2026-09-07 22:34:33 +02:00
2026-10-02 23:05:08 +02:00
2026-09-22 13:37:16 +00:00
2026-10-04 01:48:49 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 10:02:40 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 02:52:54 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 02:19:03 +02:00
2026-10-04 02:19:03 +02:00
2026-10-04 01:54:19 +02:00
2026-08-16 18:15:24 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-18 21:58:04 +02:00
2026-08-19 08:28:53 +02:00
2026-08-19 08:28:53 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 02:52:54 +02:00
2026-10-04 02:52:54 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-09-22 13:33:39 +00:00
2026-10-04 02:19:03 +02:00
2026-08-18 21:58:57 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 02:19:03 +02:00
2026-08-14 21:37:51 +00:00
2026-08-14 21:37:51 +00:00
2026-08-14 21:37:51 +00:00
2026-08-14 21:37:51 +00:00
2026-10-04 10:28:25 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 02:35:38 +02:00
2026-10-04 02:35:38 +02:00
2026-10-04 02:19:03 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 02:19:03 +02:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-10-04 02:27:54 +02:00
2026-08-20 09:31:59 +00:00
2026-10-02 23:25:04 +02:00
2026-10-02 23:25:04 +02:00
2026-10-04 09:04:23 +02:00
2026-08-20 09:31:59 +00:00
2026-09-22 13:33:39 +00:00
2026-08-20 09:31:59 +00:00
2026-10-02 23:25:04 +02:00
2026-09-22 13:33:39 +00:00
2026-10-04 02:19:03 +02:00
2026-10-04 02:27:54 +02:00
2026-10-02 23:40:38 +02:00
2026-10-04 01:31:00 +02:00
2026-10-04 08:44:46 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 10:17:08 +02:00
2026-10-02 23:04:05 +02:00
2026-10-02 23:04:05 +02:00
2026-08-14 21:33:15 +00:00
2026-10-04 01:48:49 +02:00
2026-08-18 08:35:51 +00:00
2026-08-14 21:33:15 +00:00
2026-08-30 15:47:10 +02:00
2026-10-03 23:52:31 +02:00
2026-10-02 23:13:56 +02:00
2026-08-14 21:33:15 +00:00
2026-08-16 15:33:55 +00:00
2026-08-18 22:48:11 +02:00
2026-08-18 22:48:11 +02:00
2026-08-16 15:33:55 +00:00
2026-10-02 23:05:34 +02:00
2026-10-02 23:05:34 +02:00
2026-08-20 10:32:44 +00:00
2026-08-14 21:33:15 +00:00
2026-08-14 21:33:15 +00:00
2026-08-16 20:43:47 +00:00
2026-08-15 11:35:40 +00:00
2026-09-07 22:34:33 +02:00
2026-08-16 19:45:43 +00:00
2026-08-16 19:45:43 +00:00
2026-09-22 13:37:16 +00:00
2026-10-04 02:52:54 +02:00
2026-10-02 23:21:18 +02:00
2026-09-07 22:34:33 +02:00
2026-09-07 22:34:33 +02:00
2026-09-22 13:33:39 +00:00
2026-09-07 22:34:33 +02:00
2026-09-07 22:34:57 +02:00
2026-09-07 22:34:57 +02:00
2026-09-22 13:19:37 +00:00
2026-09-22 13:19:37 +00:00
2026-08-16 15:33:55 +00:00
2026-10-02 23:13:56 +02:00
2026-10-04 02:01:11 +02:00
2026-10-04 02:01:11 +02:00