Sentence Similarity
sentence-transformers
Safetensors
multilingual
modernbert
embeddings
feature-extraction
matryoshka
retrieval
text-embeddings-inference
Instructions to use hotchpotch/bekko-embedding-v1-a8m-pt with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use hotchpotch/bekko-embedding-v1-a8m-pt with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("hotchpotch/bekko-embedding-v1-a8m-pt") sentences = [ "The weather is lovely today.", "It's so sunny outside!", "He drove to the stadium." ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] - Notebooks
- Google Colab
- Kaggle
Restore unchanged a8m-pt from 4606e9e8a489fd1e7bfe0bff703e3628d438bd17
Browse files- README.md +66 -0
- cli_args.json +0 -0
- config_sentence_transformers.json +1 -1
- eval_result.json +274 -0
- model.safetensors +1 -1
- tokenizer_config.json +2 -2
README.md
CHANGED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- multilingual
|
| 4 |
+
library_name: sentence-transformers
|
| 5 |
+
pipeline_tag: sentence-similarity
|
| 6 |
+
tags:
|
| 7 |
+
- sentence-transformers
|
| 8 |
+
- embeddings
|
| 9 |
+
- modernbert
|
| 10 |
+
- matryoshka
|
| 11 |
+
- multilingual
|
| 12 |
+
- retrieval
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
# bekko-embedding-v1-a8m-pt
|
| 16 |
+
|
| 17 |
+
Pretrained checkpoint for the bekko embedding v1 a8m release candidate. This checkpoint is the pre-fine-tuning model and is mainly useful as a base model for downstream fine-tuning or ablation work.
|
| 18 |
+
|
| 19 |
+
This upload is based on the UniR v9 QAT pretraining run:
|
| 20 |
+
|
| 21 |
+
`mmB-L4-bs8192-v1-20260301-QAT-unir_v9-full-sharedid-timing/20260504_183724/final`
|
| 22 |
+
|
| 23 |
+
## Model Details
|
| 24 |
+
|
| 25 |
+
- Architecture: ModernBERT-style encoder, 4 layers, hidden size 384
|
| 26 |
+
- Pooling: mean pooling
|
| 27 |
+
- Embedding dimension: 384
|
| 28 |
+
- Maximum sequence length: 8192 tokens
|
| 29 |
+
- Supported truncate dimensions: 384, 256, 128, 64
|
| 30 |
+
|
| 31 |
+
The model was trained with Matryoshka representation learning, so embeddings can be truncated to the supported dimensions above. Use 384 dimensions when you need the full representation, and use 256, 128, or 64 dimensions for smaller indexes or faster retrieval.
|
| 32 |
+
|
| 33 |
+
## Intended Use
|
| 34 |
+
|
| 35 |
+
This is the pre-fine-tuning checkpoint. For most retrieval use cases, prefer the fine-tuned model:
|
| 36 |
+
|
| 37 |
+
`hotchpotch/bekko-embedding-v1-a8m`
|
| 38 |
+
|
| 39 |
+
Use this pretrain checkpoint when you need to:
|
| 40 |
+
|
| 41 |
+
- reproduce or audit the fine-tuning recipe,
|
| 42 |
+
- run downstream fine-tuning from the same base,
|
| 43 |
+
- compare pretrain vs fine-tuned behavior,
|
| 44 |
+
- run ablations against the UniR v9 QAT pretrain base.
|
| 45 |
+
|
| 46 |
+
## Usage
|
| 47 |
+
|
| 48 |
+
For retrieval, prefer `encode_query()` for queries and `encode_document()` for documents. Normalize embeddings when using cosine similarity or dot-product search on normalized vectors.
|
| 49 |
+
|
| 50 |
+
```python
|
| 51 |
+
from sentence_transformers import SentenceTransformer
|
| 52 |
+
|
| 53 |
+
model = SentenceTransformer("hotchpotch/bekko-embedding-v1-a8m-pt")
|
| 54 |
+
|
| 55 |
+
queries = ["What is multilingual retrieval?"]
|
| 56 |
+
documents = ["Multilingual retrieval searches documents across languages."]
|
| 57 |
+
|
| 58 |
+
query_embeddings = model.encode_query(queries, normalize_embeddings=True)
|
| 59 |
+
document_embeddings = model.encode_document(documents, normalize_embeddings=True)
|
| 60 |
+
|
| 61 |
+
scores = query_embeddings @ document_embeddings.T
|
| 62 |
+
```
|
| 63 |
+
|
| 64 |
+
## Notes
|
| 65 |
+
|
| 66 |
+
This repository is intended for private evaluation and release-candidate validation. The model configuration in this upload is set for 8192-token inference.
|
cli_args.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
config_sentence_transformers.json
CHANGED
|
@@ -11,4 +11,4 @@
|
|
| 11 |
},
|
| 12 |
"default_prompt_name": null,
|
| 13 |
"similarity_fn_name": "cosine"
|
| 14 |
-
}
|
|
|
|
| 11 |
},
|
| 12 |
"default_prompt_name": null,
|
| 13 |
"similarity_fn_name": "cosine"
|
| 14 |
+
}
|
eval_result.json
ADDED
|
@@ -0,0 +1,274 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"nano_beir": {
|
| 3 |
+
"NanoArguAna": {
|
| 4 |
+
"MNanoArguAna-ar_cosine_ndcg@10": 0.4730338404204377,
|
| 5 |
+
"MNanoArguAna-de_cosine_ndcg@10": 0.4523858022833368,
|
| 6 |
+
"MNanoArguAna-en_cosine_ndcg@10": 0.5593102661290839,
|
| 7 |
+
"MNanoArguAna-es_cosine_ndcg@10": 0.5237835854585444,
|
| 8 |
+
"MNanoArguAna-fr_cosine_ndcg@10": 0.5168055525240536,
|
| 9 |
+
"MNanoArguAna-it_cosine_ndcg@10": 0.4727147681288612,
|
| 10 |
+
"MNanoArguAna-mean_cosine_ndcg@10": 0.49946905572549927,
|
| 11 |
+
"MNanoArguAna-no_cosine_ndcg@10": 0.5138751897026443,
|
| 12 |
+
"MNanoArguAna-pt_cosine_ndcg@10": 0.5034692240128686,
|
| 13 |
+
"MNanoArguAna-sv_cosine_ndcg@10": 0.4798432728696626
|
| 14 |
+
},
|
| 15 |
+
"NanoClimateFEVER": {
|
| 16 |
+
"MNanoClimateFEVER-ar_cosine_ndcg@10": 0.21696059280899108,
|
| 17 |
+
"MNanoClimateFEVER-de_cosine_ndcg@10": 0.28253387055871765,
|
| 18 |
+
"MNanoClimateFEVER-en_cosine_ndcg@10": 0.2916839711692681,
|
| 19 |
+
"MNanoClimateFEVER-es_cosine_ndcg@10": 0.3448162987069231,
|
| 20 |
+
"MNanoClimateFEVER-fr_cosine_ndcg@10": 0.22376330656871984,
|
| 21 |
+
"MNanoClimateFEVER-it_cosine_ndcg@10": 0.2782216799184051,
|
| 22 |
+
"MNanoClimateFEVER-mean_cosine_ndcg@10": 0.2762952444548678,
|
| 23 |
+
"MNanoClimateFEVER-no_cosine_ndcg@10": 0.27737273672960566,
|
| 24 |
+
"MNanoClimateFEVER-pt_cosine_ndcg@10": 0.29771166786184095,
|
| 25 |
+
"MNanoClimateFEVER-sv_cosine_ndcg@10": 0.27359307577133846
|
| 26 |
+
},
|
| 27 |
+
"NanoClimateFEVER-ja": {
|
| 28 |
+
"MNanoClimateFEVER-ja-mean_cosine_ndcg@10": 0.35233710769938187,
|
| 29 |
+
"MNanoClimateFEVER-ja_cosine_ndcg@10": 0.35233710769938187
|
| 30 |
+
},
|
| 31 |
+
"NanoDBPedia": {
|
| 32 |
+
"MNanoDBPedia-ar_cosine_ndcg@10": 0.49176953708703963,
|
| 33 |
+
"MNanoDBPedia-de_cosine_ndcg@10": 0.5585728125958513,
|
| 34 |
+
"MNanoDBPedia-en_cosine_ndcg@10": 0.6013656077919083,
|
| 35 |
+
"MNanoDBPedia-es_cosine_ndcg@10": 0.5635691591790805,
|
| 36 |
+
"MNanoDBPedia-fr_cosine_ndcg@10": 0.5598146452043108,
|
| 37 |
+
"MNanoDBPedia-it_cosine_ndcg@10": 0.5786410295503902,
|
| 38 |
+
"MNanoDBPedia-mean_cosine_ndcg@10": 0.5578189519284785,
|
| 39 |
+
"MNanoDBPedia-no_cosine_ndcg@10": 0.5571766709488738,
|
| 40 |
+
"MNanoDBPedia-pt_cosine_ndcg@10": 0.5449366078473445,
|
| 41 |
+
"MNanoDBPedia-sv_cosine_ndcg@10": 0.5645244971515071
|
| 42 |
+
},
|
| 43 |
+
"NanoDBPedia-ja": {
|
| 44 |
+
"MNanoDBPedia-ja-mean_cosine_ndcg@10": 0.5792733395527497,
|
| 45 |
+
"MNanoDBPedia-ja_cosine_ndcg@10": 0.5792733395527497
|
| 46 |
+
},
|
| 47 |
+
"NanoFEVER": {
|
| 48 |
+
"MNanoFEVER-ar_cosine_ndcg@10": 0.7811365282748113,
|
| 49 |
+
"MNanoFEVER-de_cosine_ndcg@10": 0.7907331358379059,
|
| 50 |
+
"MNanoFEVER-en_cosine_ndcg@10": 0.8609922099824149,
|
| 51 |
+
"MNanoFEVER-es_cosine_ndcg@10": 0.8338478791796845,
|
| 52 |
+
"MNanoFEVER-fr_cosine_ndcg@10": 0.6787773582711195,
|
| 53 |
+
"MNanoFEVER-it_cosine_ndcg@10": 0.8033124773204208,
|
| 54 |
+
"MNanoFEVER-mean_cosine_ndcg@10": 0.7977775878693328,
|
| 55 |
+
"MNanoFEVER-no_cosine_ndcg@10": 0.7903408907064573,
|
| 56 |
+
"MNanoFEVER-pt_cosine_ndcg@10": 0.8221987118236579,
|
| 57 |
+
"MNanoFEVER-sv_cosine_ndcg@10": 0.8186590994275242
|
| 58 |
+
},
|
| 59 |
+
"NanoFEVER-ja": {
|
| 60 |
+
"MNanoFEVER-ja-mean_cosine_ndcg@10": 0.724125971143851,
|
| 61 |
+
"MNanoFEVER-ja_cosine_ndcg@10": 0.724125971143851
|
| 62 |
+
},
|
| 63 |
+
"NanoFiQA2018": {
|
| 64 |
+
"MNanoFiQA2018-ar_cosine_ndcg@10": 0.32000050872745656,
|
| 65 |
+
"MNanoFiQA2018-de_cosine_ndcg@10": 0.33859658725536546,
|
| 66 |
+
"MNanoFiQA2018-en_cosine_ndcg@10": 0.45818623330557856,
|
| 67 |
+
"MNanoFiQA2018-es_cosine_ndcg@10": 0.33427309933119126,
|
| 68 |
+
"MNanoFiQA2018-fr_cosine_ndcg@10": 0.3092582897093253,
|
| 69 |
+
"MNanoFiQA2018-it_cosine_ndcg@10": 0.33185375268271744,
|
| 70 |
+
"MNanoFiQA2018-mean_cosine_ndcg@10": 0.3243882573066615,
|
| 71 |
+
"MNanoFiQA2018-no_cosine_ndcg@10": 0.29262339645711927,
|
| 72 |
+
"MNanoFiQA2018-pt_cosine_ndcg@10": 0.2682571114141015,
|
| 73 |
+
"MNanoFiQA2018-sv_cosine_ndcg@10": 0.2664453368770985
|
| 74 |
+
},
|
| 75 |
+
"NanoFiQA2018-ja": {
|
| 76 |
+
"MNanoFiQA2018-ja-mean_cosine_ndcg@10": 0.3222650200367273,
|
| 77 |
+
"MNanoFiQA2018-ja_cosine_ndcg@10": 0.3222650200367273
|
| 78 |
+
},
|
| 79 |
+
"NanoHotpotQA": {
|
| 80 |
+
"MNanoHotpotQA-ar_cosine_ndcg@10": 0.6956427737846929,
|
| 81 |
+
"MNanoHotpotQA-de_cosine_ndcg@10": 0.7701008639893859,
|
| 82 |
+
"MNanoHotpotQA-en_cosine_ndcg@10": 0.7690504439859169,
|
| 83 |
+
"MNanoHotpotQA-es_cosine_ndcg@10": 0.722497873994412,
|
| 84 |
+
"MNanoHotpotQA-fr_cosine_ndcg@10": 0.7463668087829665,
|
| 85 |
+
"MNanoHotpotQA-it_cosine_ndcg@10": 0.7464549361636561,
|
| 86 |
+
"MNanoHotpotQA-mean_cosine_ndcg@10": 0.7416132572210601,
|
| 87 |
+
"MNanoHotpotQA-no_cosine_ndcg@10": 0.7401936856204756,
|
| 88 |
+
"MNanoHotpotQA-pt_cosine_ndcg@10": 0.734274156218506,
|
| 89 |
+
"MNanoHotpotQA-sv_cosine_ndcg@10": 0.7499377724495281
|
| 90 |
+
},
|
| 91 |
+
"NanoHotpotQA-ja": {
|
| 92 |
+
"MNanoHotpotQA-ja-mean_cosine_ndcg@10": 0.6701411332471646,
|
| 93 |
+
"MNanoHotpotQA-ja_cosine_ndcg@10": 0.6701411332471646
|
| 94 |
+
},
|
| 95 |
+
"NanoMSMARCO": {
|
| 96 |
+
"MNanoMSMARCO-ar_cosine_ndcg@10": 0.43785211849573136,
|
| 97 |
+
"MNanoMSMARCO-de_cosine_ndcg@10": 0.47974421951438656,
|
| 98 |
+
"MNanoMSMARCO-en_cosine_ndcg@10": 0.5476519075374822,
|
| 99 |
+
"MNanoMSMARCO-es_cosine_ndcg@10": 0.49934336391420997,
|
| 100 |
+
"MNanoMSMARCO-fr_cosine_ndcg@10": 0.4821488420046777,
|
| 101 |
+
"MNanoMSMARCO-it_cosine_ndcg@10": 0.5089587403780814,
|
| 102 |
+
"MNanoMSMARCO-mean_cosine_ndcg@10": 0.4732463048264413,
|
| 103 |
+
"MNanoMSMARCO-no_cosine_ndcg@10": 0.4077398253644848,
|
| 104 |
+
"MNanoMSMARCO-pt_cosine_ndcg@10": 0.49901980528206363,
|
| 105 |
+
"MNanoMSMARCO-sv_cosine_ndcg@10": 0.39675792094685436
|
| 106 |
+
},
|
| 107 |
+
"NanoMSMARCO-ja": {
|
| 108 |
+
"MNanoMSMARCO-ja-mean_cosine_ndcg@10": 0.4945060711606406,
|
| 109 |
+
"MNanoMSMARCO-ja_cosine_ndcg@10": 0.4945060711606406
|
| 110 |
+
},
|
| 111 |
+
"NanoNFCorpus": {
|
| 112 |
+
"MNanoNFCorpus-ar_cosine_ndcg@10": 0.22716153322345686,
|
| 113 |
+
"MNanoNFCorpus-de_cosine_ndcg@10": 0.24481731941369972,
|
| 114 |
+
"MNanoNFCorpus-en_cosine_ndcg@10": 0.33595088460224604,
|
| 115 |
+
"MNanoNFCorpus-es_cosine_ndcg@10": 0.30753131265305245,
|
| 116 |
+
"MNanoNFCorpus-fr_cosine_ndcg@10": 0.3188369766309311,
|
| 117 |
+
"MNanoNFCorpus-it_cosine_ndcg@10": 0.28557450822422525,
|
| 118 |
+
"MNanoNFCorpus-mean_cosine_ndcg@10": 0.28659175055370834,
|
| 119 |
+
"MNanoNFCorpus-no_cosine_ndcg@10": 0.2725952012643174,
|
| 120 |
+
"MNanoNFCorpus-pt_cosine_ndcg@10": 0.2979627431402395,
|
| 121 |
+
"MNanoNFCorpus-sv_cosine_ndcg@10": 0.2888952758312067
|
| 122 |
+
},
|
| 123 |
+
"NanoNFCorpus-ja": {
|
| 124 |
+
"MNanoNFCorpus-ja-mean_cosine_ndcg@10": 0.2799070812314455,
|
| 125 |
+
"MNanoNFCorpus-ja_cosine_ndcg@10": 0.2799070812314455
|
| 126 |
+
},
|
| 127 |
+
"NanoNQ": {
|
| 128 |
+
"MNanoNQ-ar_cosine_ndcg@10": 0.45515849206895986,
|
| 129 |
+
"MNanoNQ-de_cosine_ndcg@10": 0.49779262972522065,
|
| 130 |
+
"MNanoNQ-en_cosine_ndcg@10": 0.6090497523222711,
|
| 131 |
+
"MNanoNQ-es_cosine_ndcg@10": 0.43973148585510474,
|
| 132 |
+
"MNanoNQ-fr_cosine_ndcg@10": 0.5161833937286435,
|
| 133 |
+
"MNanoNQ-it_cosine_ndcg@10": 0.4683263144521548,
|
| 134 |
+
"MNanoNQ-mean_cosine_ndcg@10": 0.47901010889360535,
|
| 135 |
+
"MNanoNQ-no_cosine_ndcg@10": 0.44336355209988704,
|
| 136 |
+
"MNanoNQ-pt_cosine_ndcg@10": 0.4340017129052539,
|
| 137 |
+
"MNanoNQ-sv_cosine_ndcg@10": 0.44748364688495246
|
| 138 |
+
},
|
| 139 |
+
"NanoNQ-ja": {
|
| 140 |
+
"MNanoNQ-ja-mean_cosine_ndcg@10": 0.5403929314434924,
|
| 141 |
+
"MNanoNQ-ja_cosine_ndcg@10": 0.5403929314434924
|
| 142 |
+
},
|
| 143 |
+
"NanoQuoraRetrieval": {
|
| 144 |
+
"MNanoQuoraRetrieval-ar_cosine_ndcg@10": 0.8180530918417213,
|
| 145 |
+
"MNanoQuoraRetrieval-de_cosine_ndcg@10": 0.8356060159578926,
|
| 146 |
+
"MNanoQuoraRetrieval-en_cosine_ndcg@10": 0.9346258899873209,
|
| 147 |
+
"MNanoQuoraRetrieval-es_cosine_ndcg@10": 0.8821377839507355,
|
| 148 |
+
"MNanoQuoraRetrieval-fr_cosine_ndcg@10": 0.7925711224415832,
|
| 149 |
+
"MNanoQuoraRetrieval-it_cosine_ndcg@10": 0.8340771342434293,
|
| 150 |
+
"MNanoQuoraRetrieval-mean_cosine_ndcg@10": 0.8307165948334699,
|
| 151 |
+
"MNanoQuoraRetrieval-no_cosine_ndcg@10": 0.7588980586522337,
|
| 152 |
+
"MNanoQuoraRetrieval-pt_cosine_ndcg@10": 0.8207898164276353,
|
| 153 |
+
"MNanoQuoraRetrieval-sv_cosine_ndcg@10": 0.7996904399986784
|
| 154 |
+
},
|
| 155 |
+
"NanoQuoraRetrieval-ja": {
|
| 156 |
+
"MNanoQuoraRetrieval-ja-mean_cosine_ndcg@10": 0.8935715249766721,
|
| 157 |
+
"MNanoQuoraRetrieval-ja_cosine_ndcg@10": 0.8935715249766721
|
| 158 |
+
},
|
| 159 |
+
"NanoSCIDOCS": {
|
| 160 |
+
"MNanoSCIDOCS-ar_cosine_ndcg@10": 0.2935560321306876,
|
| 161 |
+
"MNanoSCIDOCS-de_cosine_ndcg@10": 0.30095111574851746,
|
| 162 |
+
"MNanoSCIDOCS-en_cosine_ndcg@10": 0.3821570242399539,
|
| 163 |
+
"MNanoSCIDOCS-es_cosine_ndcg@10": 0.3421588969001715,
|
| 164 |
+
"MNanoSCIDOCS-fr_cosine_ndcg@10": 0.3082776916852559,
|
| 165 |
+
"MNanoSCIDOCS-it_cosine_ndcg@10": 0.3109347050091602,
|
| 166 |
+
"MNanoSCIDOCS-mean_cosine_ndcg@10": 0.31933584744766974,
|
| 167 |
+
"MNanoSCIDOCS-no_cosine_ndcg@10": 0.2924377548797874,
|
| 168 |
+
"MNanoSCIDOCS-pt_cosine_ndcg@10": 0.3284645551969043,
|
| 169 |
+
"MNanoSCIDOCS-sv_cosine_ndcg@10": 0.31508485123858976
|
| 170 |
+
},
|
| 171 |
+
"NanoSCIDOCS-ja": {
|
| 172 |
+
"MNanoSCIDOCS-ja-mean_cosine_ndcg@10": 0.34071630056860536,
|
| 173 |
+
"MNanoSCIDOCS-ja_cosine_ndcg@10": 0.34071630056860536
|
| 174 |
+
},
|
| 175 |
+
"NanoSciFact": {
|
| 176 |
+
"MNanoSciFact-ar_cosine_ndcg@10": 0.5288426430511044,
|
| 177 |
+
"MNanoSciFact-de_cosine_ndcg@10": 0.6622217521928413,
|
| 178 |
+
"MNanoSciFact-en_cosine_ndcg@10": 0.7628440491657529,
|
| 179 |
+
"MNanoSciFact-es_cosine_ndcg@10": 0.6527242947213842,
|
| 180 |
+
"MNanoSciFact-fr_cosine_ndcg@10": 0.676593207849287,
|
| 181 |
+
"MNanoSciFact-it_cosine_ndcg@10": 0.6419598857336184,
|
| 182 |
+
"MNanoSciFact-mean_cosine_ndcg@10": 0.6344163638808986,
|
| 183 |
+
"MNanoSciFact-no_cosine_ndcg@10": 0.6071069309322736,
|
| 184 |
+
"MNanoSciFact-pt_cosine_ndcg@10": 0.5776104964850115,
|
| 185 |
+
"MNanoSciFact-sv_cosine_ndcg@10": 0.5998440147968133
|
| 186 |
+
},
|
| 187 |
+
"NanoTouche2020": {
|
| 188 |
+
"MNanoTouche2020-ar_cosine_ndcg@10": 0.4081229095460993,
|
| 189 |
+
"MNanoTouche2020-de_cosine_ndcg@10": 0.4388518349118278,
|
| 190 |
+
"MNanoTouche2020-en_cosine_ndcg@10": 0.5124116336684273,
|
| 191 |
+
"MNanoTouche2020-es_cosine_ndcg@10": 0.44235774385322657,
|
| 192 |
+
"MNanoTouche2020-fr_cosine_ndcg@10": 0.3992503328132656,
|
| 193 |
+
"MNanoTouche2020-it_cosine_ndcg@10": 0.4076703520501428,
|
| 194 |
+
"MNanoTouche2020-mean_cosine_ndcg@10": 0.4335759729678913,
|
| 195 |
+
"MNanoTouche2020-no_cosine_ndcg@10": 0.4241551560510449,
|
| 196 |
+
"MNanoTouche2020-pt_cosine_ndcg@10": 0.43129192978542286,
|
| 197 |
+
"MNanoTouche2020-sv_cosine_ndcg@10": 0.43807186403156484
|
| 198 |
+
}
|
| 199 |
+
},
|
| 200 |
+
"nano_fineweb2ir": {
|
| 201 |
+
"NanoFineWeb2IR-arb_Arab_cosine_ndcg@10": 0.9438507213043262,
|
| 202 |
+
"NanoFineWeb2IR-ces_Latn_cosine_ndcg@10": 0.9452371901428583,
|
| 203 |
+
"NanoFineWeb2IR-cmn_Hani_cosine_ndcg@10": 0.9238507213043262,
|
| 204 |
+
"NanoFineWeb2IR-dan_Latn_cosine_ndcg@10": 0.9578557852142875,
|
| 205 |
+
"NanoFineWeb2IR-deu_Latn_cosine_ndcg@10": 0.9926185950714291,
|
| 206 |
+
"NanoFineWeb2IR-ell_Grek_cosine_ndcg@10": 0.9406803948112681,
|
| 207 |
+
"NanoFineWeb2IR-eng_Latn_cosine_ndcg@10": 0.9686135311614679,
|
| 208 |
+
"NanoFineWeb2IR-fas_Arab_cosine_ndcg@10": 0.9287119211911774,
|
| 209 |
+
"NanoFineWeb2IR-fra_Latn_cosine_ndcg@10": 0.8653058616513755,
|
| 210 |
+
"NanoFineWeb2IR-hun_Latn_cosine_ndcg@10": 0.9332476622362154,
|
| 211 |
+
"NanoFineWeb2IR-ind_Latn_cosine_ndcg@10": 0.9200983900297096,
|
| 212 |
+
"NanoFineWeb2IR-ita_Latn_cosine_ndcg@10": 0.8565362699694403,
|
| 213 |
+
"NanoFineWeb2IR-jpn_Jpan_cosine_ndcg@10": 0.9926185950714291,
|
| 214 |
+
"NanoFineWeb2IR-nld_Latn_cosine_ndcg@10": 0.9426185950714292,
|
| 215 |
+
"NanoFineWeb2IR-pol_Latn_cosine_ndcg@10": 0.9278557852142875,
|
| 216 |
+
"NanoFineWeb2IR-por_Latn_cosine_ndcg@10": 0.9478557852142875,
|
| 217 |
+
"NanoFineWeb2IR-rus_Cyrl_cosine_ndcg@10": 0.96,
|
| 218 |
+
"NanoFineWeb2IR-spa_Latn_cosine_ndcg@10": 0.9483998915977869,
|
| 219 |
+
"NanoFineWeb2IR-swe_Latn_cosine_ndcg@10": 0.9852371901428583,
|
| 220 |
+
"NanoFineWeb2IR-tur_Latn_cosine_ndcg@10": 0.9429742462875491,
|
| 221 |
+
"NanoFineWeb2IR-vie_Latn_cosine_ndcg@10": 0.9167836778214311,
|
| 222 |
+
"NanoFineWeb2IR_mean_cosine_ndcg@10": 0.9400452766909018
|
| 223 |
+
},
|
| 224 |
+
"nano_miracl": {
|
| 225 |
+
"NanoMIRACL-ar_cosine_ndcg@10": 0.6955587747416526,
|
| 226 |
+
"NanoMIRACL-bn_cosine_ndcg@10": 0.6443383048804079,
|
| 227 |
+
"NanoMIRACL-de_cosine_ndcg@10": 0.6099802384119607,
|
| 228 |
+
"NanoMIRACL-en_cosine_ndcg@10": 0.6734205400931867,
|
| 229 |
+
"NanoMIRACL-es_cosine_ndcg@10": 0.6876805333372731,
|
| 230 |
+
"NanoMIRACL-fa_cosine_ndcg@10": 0.6484924725291072,
|
| 231 |
+
"NanoMIRACL-fi_cosine_ndcg@10": 0.7461712879088102,
|
| 232 |
+
"NanoMIRACL-fr_cosine_ndcg@10": 0.6217460185241914,
|
| 233 |
+
"NanoMIRACL-hi_cosine_ndcg@10": 0.6157970155676352,
|
| 234 |
+
"NanoMIRACL-id_cosine_ndcg@10": 0.656938278826782,
|
| 235 |
+
"NanoMIRACL-ja_cosine_ndcg@10": 0.6269369051853845,
|
| 236 |
+
"NanoMIRACL-ko_cosine_ndcg@10": 0.5944354127162638,
|
| 237 |
+
"NanoMIRACL-ru_cosine_ndcg@10": 0.7159646473373542,
|
| 238 |
+
"NanoMIRACL-sw_cosine_ndcg@10": 0.5656108922525411,
|
| 239 |
+
"NanoMIRACL-te_cosine_ndcg@10": 0.9386391949847088,
|
| 240 |
+
"NanoMIRACL-th_cosine_ndcg@10": 0.7792507958853974,
|
| 241 |
+
"NanoMIRACL-yo_cosine_ndcg@10": 0.6907384653898725,
|
| 242 |
+
"NanoMIRACL-zh_cosine_ndcg@10": 0.6344802314104463,
|
| 243 |
+
"NanoMIRACL_mean_cosine_ndcg@10": 0.6747877783323876
|
| 244 |
+
},
|
| 245 |
+
"summary": {
|
| 246 |
+
"NanoArguAna_mean_ndcg@10": 0.49946905572549927,
|
| 247 |
+
"NanoClimateFEVER-ja_mean_ndcg@10": 0.35233710769938187,
|
| 248 |
+
"NanoClimateFEVER_mean_ndcg@10": 0.2762952444548678,
|
| 249 |
+
"NanoDBPedia-ja_mean_ndcg@10": 0.5792733395527497,
|
| 250 |
+
"NanoDBPedia_mean_ndcg@10": 0.5578189519284785,
|
| 251 |
+
"NanoFEVER-ja_mean_ndcg@10": 0.724125971143851,
|
| 252 |
+
"NanoFEVER_mean_ndcg@10": 0.7977775878693328,
|
| 253 |
+
"NanoFiQA2018-ja_mean_ndcg@10": 0.3222650200367273,
|
| 254 |
+
"NanoFiQA2018_mean_ndcg@10": 0.3243882573066615,
|
| 255 |
+
"NanoHotpotQA-ja_mean_ndcg@10": 0.6701411332471646,
|
| 256 |
+
"NanoHotpotQA_mean_ndcg@10": 0.7416132572210601,
|
| 257 |
+
"NanoMSMARCO-ja_mean_ndcg@10": 0.4945060711606406,
|
| 258 |
+
"NanoMSMARCO_mean_ndcg@10": 0.4732463048264413,
|
| 259 |
+
"NanoNFCorpus-ja_mean_ndcg@10": 0.2799070812314455,
|
| 260 |
+
"NanoNFCorpus_mean_ndcg@10": 0.28659175055370834,
|
| 261 |
+
"NanoNQ-ja_mean_ndcg@10": 0.5403929314434924,
|
| 262 |
+
"NanoNQ_mean_ndcg@10": 0.47901010889360535,
|
| 263 |
+
"NanoQuoraRetrieval-ja_mean_ndcg@10": 0.8935715249766721,
|
| 264 |
+
"NanoQuoraRetrieval_mean_ndcg@10": 0.8307165948334699,
|
| 265 |
+
"NanoSCIDOCS-ja_mean_ndcg@10": 0.34071630056860536,
|
| 266 |
+
"NanoSCIDOCS_mean_ndcg@10": 0.31933584744766974,
|
| 267 |
+
"NanoSciFact_mean_ndcg@10": 0.6344163638808986,
|
| 268 |
+
"NanoTouche2020_mean_ndcg@10": 0.4335759729678913,
|
| 269 |
+
"all_mean_ndcg@10": 0.5386529933597441,
|
| 270 |
+
"nano_fineweb2ir_mean_ndcg@10": 0.9400452766909018,
|
| 271 |
+
"nano_miracl_mean_ndcg@10": 0.6747877783323876
|
| 272 |
+
},
|
| 273 |
+
"timestamp_utc": "2026-07-06T18:00:23.553943+00:00"
|
| 274 |
+
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 211952896
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b474c13e826f1652bffef917e1e3a1dccb172607087ccd0d5a0a61dc91c83cc7
|
| 3 |
size 211952896
|
tokenizer_config.json
CHANGED
|
@@ -2004,12 +2004,12 @@
|
|
| 2004 |
"eos_token": "<eos>",
|
| 2005 |
"extra_special_tokens": {},
|
| 2006 |
"mask_token": "<mask>",
|
| 2007 |
-
"max_length":
|
| 2008 |
"model_input_names": [
|
| 2009 |
"input_ids",
|
| 2010 |
"attention_mask"
|
| 2011 |
],
|
| 2012 |
-
"model_max_length":
|
| 2013 |
"pad_to_multiple_of": null,
|
| 2014 |
"pad_token": "<pad>",
|
| 2015 |
"pad_token_type_id": 0,
|
|
|
|
| 2004 |
"eos_token": "<eos>",
|
| 2005 |
"extra_special_tokens": {},
|
| 2006 |
"mask_token": "<mask>",
|
| 2007 |
+
"max_length": 8192,
|
| 2008 |
"model_input_names": [
|
| 2009 |
"input_ids",
|
| 2010 |
"attention_mask"
|
| 2011 |
],
|
| 2012 |
+
"model_max_length": 8192,
|
| 2013 |
"pad_to_multiple_of": null,
|
| 2014 |
"pad_token": "<pad>",
|
| 2015 |
"pad_token_type_id": 0,
|