Monitoring des modèles
Meilleurs modèles (RRF, classés par nDCG@10) · 2026-W46
| Modèle | Version | Recall@10 | MRR | nDCG@10 | Run |
|---|---|---|---|---|---|
| google/embeddinggemma-300m | 0.178 | 0.383 | 0.189 | 2026-W46-gemma | |
| Qwen/Qwen3-Embedding-0.6B | 0.177 | 0.390 | 0.188 | 2026-W46-qwen3 | |
| jinaai/jina-embeddings-v3 | 0.169 | 0.393 | 0.184 | 2026-W46-jinav3 |
nDCG@10 par modèle et méthode · 2026-W46
Évolution des métriques
Modèles
Une couleur par modèle, un style de trait par métrique.
Résultats détaillés
| Run | Modèle | Jeu | BM25 | E5 | RRF | Config | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| R@10 | MRR | nDCG | R@10 | MRR | nDCG | R@10 | MRR | nDCG | ||||
| 2026-W46-testbaseE5 | intfloat/multilingual-e5-base | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.149 | 0.347 | 0.160 | 0.173 | 0.370 | 0.183 | – |
| 2026-W46-testsmalle5 | intfloat/multilingual-e5-small | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.149 | 0.338 | 0.159 | 0.164 | 0.367 | 0.175 | – |
| 2026-W46-e5basetest | intfloat/multilingual-e5-base | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.149 | 0.347 | 0.160 | 0.173 | 0.370 | 0.183 | – |
| 2026-W46-gemma | google/embeddinggemma-300m | v1-2ecf814e | 0.156 | 0.332 | 0.160 | 0.149 | 0.383 | 0.166 | 0.178 | 0.383 | 0.189 | – |
| 2026-W46-qwen3 | Qwen/Qwen3-Embedding-0.6B | v1-2ecf814e | 0.156 | 0.335 | 0.161 | 0.150 | 0.351 | 0.161 | 0.177 | 0.390 | 0.188 | – |
| 2026-W46-nomic | nomic-ai/nomic-embed-text-v2-moe | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.137 | 0.334 | 0.147 | 0.169 | 0.363 | 0.177 | – |
| 2026-W46-jinav3 | jinaai/jina-embeddings-v3 | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.129 | 0.327 | 0.143 | 0.169 | 0.393 | 0.184 | – |
| 2026-W46-gte | Alibaba-NLP/gte-multilingual-base | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.129 | 0.306 | 0.137 | 0.168 | 0.360 | 0.178 | – |
| 2026-W46-granite | ibm-granite/granite-embedding-278m-multilingual | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.130 | 0.324 | 0.140 | 0.168 | 0.367 | 0.179 | – |
| 2026-W46-e5small | intfloat/multilingual-e5-small | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.149 | 0.338 | 0.159 | 0.164 | 0.367 | 0.175 | – |
| 2026-W46-e5large | intfloat/multilingual-e5-large-instruct | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.144 | 0.398 | 0.166 | 0.168 | 0.362 | 0.177 | – |
| 2026-W46-e5base | intfloat/multilingual-e5-base | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.149 | 0.347 | 0.160 | 0.173 | 0.370 | 0.183 | – |
| 2026-W46-bgem3 | BAAI/bge-m3 | v1-2ecf814e | 0.155 | 0.330 | 0.159 | 0.135 | 0.326 | 0.144 | 0.168 | 0.349 | 0.174 | – |
Historique des runs
| Run | Cluster | Import | Fin | Durée | Mode | Erreur | Config |
|---|---|---|---|---|---|---|---|
| 2026-W46-gemma | failed | not_applicable | 20/11/2026 18:49 | 69 min | import | SystemExit: SIGTERM reçu (limite de temps ou scancel) | voir{
"batch_size": 256,
"chunking_version": "c1",
"default_tag": "gemma",
"dtype": "float32",
"embedding_dim": 768,
"max_tokens": 512,
"model_name": "google/embeddinggemma-300m",
"model_version": "embeddinggemma-57c266a7",
"normalize": true,
"passage_prefix": "title: none | text: ",
"query_prefix": "task: search result | query: ",
"revision": "57c266a740f537b4dc058e1b0cda161fd15afa75",
"trust_remote_code": false
} |
| 2026-W46-testbaseE5 | success | not_applicable | 20/11/2026 09:51 | 18 min | analysis | – | |
| 2026-W46-testsmalle5 | success | not_applicable | 20/11/2026 09:17 | 12 min | analysis | – | |
| 2026-W46-e5basetest | success | not_applicable | 19/11/2026 21:51 | 18 min | analysis | – | |
| 2026-W46-qwen3 | failed | not_applicable | 18/11/2026 02:49 | 1 min | import | ValueError: The checkpoint you are trying to load has model type `qwen3` but Transformers does not recognize this architecture. This could be because of an issue with the checkpoint, or because your version of Transformers is out of date. You can update Transformers with the command `pip install --upgrade transformers`. If this does not work, and the checkpoint is very new, then there may not be a release version that supports this model yet. In this case, you can get the most up-to-date code b | – |
| 2026-W46-qwen3 | success | not_applicable | 18/11/2026 09:16 | 96 min | analysis | – | |
| 2026-W46-nomic | success | not_applicable | 18/11/2026 02:49 | 36 min | analysis | – | |
| 2026-W46-jinav3 | success | not_applicable | 18/11/2026 02:13 | 65 min | analysis | – | |
| 2026-W46-gte | success | not_applicable | 18/11/2026 01:08 | 23 min | analysis | – | |
| 2026-W46-granite | success | not_applicable | 18/11/2026 00:44 | 17 min | analysis | – | |
| 2026-W46-gemma | success | not_applicable | 18/11/2026 15:05 | 95 min | analysis | – | |
| 2026-W46-e5small | success | not_applicable | 18/11/2026 00:27 | 11 min | analysis | – | |
| 2026-W46-e5large | success | not_applicable | 18/11/2026 00:16 | 39 min | analysis | – | |
| 2026-W46-e5base | success | not_applicable | 17/11/2026 23:37 | 17 min | analysis | – | |
| 2026-W46-bgem3 | success | not_applicable | 17/11/2026 23:19 | 39 min | analysis | – |