multilingual-e5-small-en-pruned
This model is a token-embedding pruned version of intfloat/multilingual-e5-small.
Token-embedding pruning clusters semantically similar tokens in the embedding space (using DBSCAN) and merges each cluster into a single shared embedding, shrinking the vocabulary and reducing memory without retraining the transformer layers.
How to use
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("jangedoo/multilingual-e5-small-en-pruned",
trust_remote_code=True)
embeddings = model.encode(["Hello world", "How are you?"])
Note:
trust_remote_code=Trueis required because the model ships a small custom tokenizer class (pruned_tokenizer.py) that applies the id remapping after tokenization. No additional package installation is needed.
Pruning statistics
| Base | Pruned | Reduction | |
|---|---|---|---|
| Vocab size | 250,037 | 71,547 | 71.39% |
| Total parameters | 117,653,760 | 49,113,600 | 58.26% |
| Embedding parameters | 96,014,208 | 27,474,048 | 71.39% |
| Embedding size (MB) | 366.3 | 104.8 | 261.5 MB saved |
Evaluation
| Dataset / Metric | Base | Pruned | Relative (base = 1.0) |
|---|---|---|---|
| stsb / stsb_pearson_cosine | 0.8092 | 0.8091 | 1.0000 |
| stsb / stsb_spearman_cosine | 0.8359 | 0.8359 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_accuracy@1 | 0.3000 | 0.3000 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_accuracy@3 | 0.4200 | 0.4200 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_accuracy@5 | 0.5000 | 0.5000 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_accuracy@10 | 0.6600 | 0.6600 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_precision@1 | 0.3000 | 0.3000 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_precision@3 | 0.1533 | 0.1533 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_precision@5 | 0.1160 | 0.1160 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_precision@10 | 0.0880 | 0.0880 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_recall@1 | 0.1500 | 0.1500 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_recall@3 | 0.2000 | 0.2000 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_recall@5 | 0.2433 | 0.2433 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_recall@10 | 0.3530 | 0.3530 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_ndcg@10 | 0.2927 | 0.2927 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_mrr@10 | 0.3906 | 0.3906 | 1.0000 |
| nanobeir / NanoClimateFEVER_cosine_map@100 | 0.2358 | 0.2358 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_accuracy@1 | 0.5800 | 0.5600 | 0.9655 |
| nanobeir / NanoDBPedia_cosine_accuracy@3 | 0.8400 | 0.8400 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_accuracy@5 | 0.8800 | 0.8800 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_accuracy@10 | 0.9600 | 0.9600 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_precision@1 | 0.5800 | 0.5600 | 0.9655 |
| nanobeir / NanoDBPedia_cosine_precision@3 | 0.5400 | 0.5400 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_precision@5 | 0.5200 | 0.5200 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_precision@10 | 0.4300 | 0.4320 | 1.0047 |
| nanobeir / NanoDBPedia_cosine_recall@1 | 0.0755 | 0.0730 | 0.9669 |
| nanobeir / NanoDBPedia_cosine_recall@3 | 0.1534 | 0.1534 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_recall@5 | 0.2049 | 0.2049 | 1.0000 |
| nanobeir / NanoDBPedia_cosine_recall@10 | 0.3126 | 0.3135 | 1.0028 |
| nanobeir / NanoDBPedia_cosine_ndcg@10 | 0.5371 | 0.5368 | 0.9994 |
| nanobeir / NanoDBPedia_cosine_mrr@10 | 0.7175 | 0.7075 | 0.9861 |
| nanobeir / NanoDBPedia_cosine_map@100 | 0.3988 | 0.3975 | 0.9967 |
| nanobeir / NanoFEVER_cosine_accuracy@1 | 0.6200 | 0.6200 | 1.0000 |
| nanobeir / NanoFEVER_cosine_accuracy@3 | 0.8800 | 0.8800 | 1.0000 |
| nanobeir / NanoFEVER_cosine_accuracy@5 | 0.9400 | 0.9400 | 1.0000 |
| nanobeir / NanoFEVER_cosine_accuracy@10 | 0.9800 | 0.9800 | 1.0000 |
| nanobeir / NanoFEVER_cosine_precision@1 | 0.6200 | 0.6200 | 1.0000 |
| nanobeir / NanoFEVER_cosine_precision@3 | 0.3000 | 0.3000 | 1.0000 |
| nanobeir / NanoFEVER_cosine_precision@5 | 0.1960 | 0.1960 | 1.0000 |
| nanobeir / NanoFEVER_cosine_precision@10 | 0.1020 | 0.1020 | 1.0000 |
| nanobeir / NanoFEVER_cosine_recall@1 | 0.5867 | 0.5867 | 1.0000 |
| nanobeir / NanoFEVER_cosine_recall@3 | 0.8433 | 0.8433 | 1.0000 |
| nanobeir / NanoFEVER_cosine_recall@5 | 0.9033 | 0.9033 | 1.0000 |
| nanobeir / NanoFEVER_cosine_recall@10 | 0.9333 | 0.9333 | 1.0000 |
| nanobeir / NanoFEVER_cosine_ndcg@10 | 0.7897 | 0.7897 | 1.0000 |
| nanobeir / NanoFEVER_cosine_mrr@10 | 0.7592 | 0.7592 | 1.0000 |
| nanobeir / NanoFEVER_cosine_map@100 | 0.7338 | 0.7338 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_accuracy@1 | 0.3600 | 0.3600 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_accuracy@3 | 0.5600 | 0.5600 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_accuracy@5 | 0.6200 | 0.6200 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_accuracy@10 | 0.6600 | 0.6600 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_precision@1 | 0.3600 | 0.3600 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_precision@3 | 0.2400 | 0.2400 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_precision@5 | 0.1800 | 0.1800 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_precision@10 | 0.1060 | 0.1060 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_recall@1 | 0.1801 | 0.1801 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_recall@3 | 0.3545 | 0.3545 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_recall@5 | 0.4403 | 0.4403 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_recall@10 | 0.4878 | 0.4878 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_ndcg@10 | 0.3956 | 0.3956 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_mrr@10 | 0.4630 | 0.4630 | 1.0000 |
| nanobeir / NanoFiQA2018_cosine_map@100 | 0.3380 | 0.3380 | 1.0002 |
| nanobeir / NanoHotpotQA_cosine_accuracy@1 | 0.7800 | 0.7800 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_accuracy@3 | 0.9200 | 0.9200 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_accuracy@5 | 0.9600 | 0.9600 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_accuracy@10 | 0.9800 | 0.9800 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_precision@1 | 0.7800 | 0.7800 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_precision@3 | 0.5000 | 0.5000 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_precision@5 | 0.3240 | 0.3240 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_precision@10 | 0.1720 | 0.1720 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_recall@1 | 0.3900 | 0.3900 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_recall@3 | 0.7500 | 0.7500 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_recall@5 | 0.8100 | 0.8100 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_recall@10 | 0.8600 | 0.8600 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_ndcg@10 | 0.7997 | 0.7997 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_mrr@10 | 0.8600 | 0.8600 | 1.0000 |
| nanobeir / NanoHotpotQA_cosine_map@100 | 0.7435 | 0.7435 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_accuracy@1 | 0.4200 | 0.4200 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_accuracy@3 | 0.5800 | 0.6000 | 1.0345 |
| nanobeir / NanoMSMARCO_cosine_accuracy@5 | 0.7600 | 0.7600 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_accuracy@10 | 0.8600 | 0.8600 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_precision@1 | 0.4200 | 0.4200 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_precision@3 | 0.1933 | 0.2000 | 1.0345 |
| nanobeir / NanoMSMARCO_cosine_precision@5 | 0.1520 | 0.1520 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_precision@10 | 0.0860 | 0.0860 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_recall@1 | 0.4200 | 0.4200 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_recall@3 | 0.5800 | 0.6000 | 1.0345 |
| nanobeir / NanoMSMARCO_cosine_recall@5 | 0.7600 | 0.7600 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_recall@10 | 0.8600 | 0.8600 | 1.0000 |
| nanobeir / NanoMSMARCO_cosine_ndcg@10 | 0.6187 | 0.6210 | 1.0037 |
| nanobeir / NanoMSMARCO_cosine_mrr@10 | 0.5436 | 0.5463 | 1.0049 |
| nanobeir / NanoMSMARCO_cosine_map@100 | 0.5517 | 0.5543 | 1.0048 |
| nanobeir / NanoNFCorpus_cosine_accuracy@1 | 0.4200 | 0.4200 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_accuracy@3 | 0.5000 | 0.5000 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_accuracy@5 | 0.5600 | 0.5600 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_accuracy@10 | 0.6400 | 0.6400 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_precision@1 | 0.4200 | 0.4200 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_precision@3 | 0.3267 | 0.3267 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_precision@5 | 0.3280 | 0.3280 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_precision@10 | 0.2520 | 0.2520 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_recall@1 | 0.0148 | 0.0148 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_recall@3 | 0.0442 | 0.0442 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_recall@5 | 0.0772 | 0.0772 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_recall@10 | 0.0999 | 0.0999 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_ndcg@10 | 0.2937 | 0.2937 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_mrr@10 | 0.4829 | 0.4829 | 1.0000 |
| nanobeir / NanoNFCorpus_cosine_map@100 | 0.1046 | 0.1047 | 1.0009 |
| nanobeir / NanoNQ_cosine_accuracy@1 | 0.5400 | 0.4600 | 0.8519 |
| nanobeir / NanoNQ_cosine_accuracy@3 | 0.6400 | 0.6000 | 0.9375 |
| nanobeir / NanoNQ_cosine_accuracy@5 | 0.7000 | 0.6800 | 0.9714 |
| nanobeir / NanoNQ_cosine_accuracy@10 | 0.8200 | 0.8000 | 0.9756 |
| nanobeir / NanoNQ_cosine_precision@1 | 0.5400 | 0.4600 | 0.8519 |
| nanobeir / NanoNQ_cosine_precision@3 | 0.2133 | 0.2067 | 0.9688 |
| nanobeir / NanoNQ_cosine_precision@5 | 0.1480 | 0.1440 | 0.9730 |
| nanobeir / NanoNQ_cosine_precision@10 | 0.0900 | 0.0880 | 0.9778 |
| nanobeir / NanoNQ_cosine_recall@1 | 0.4900 | 0.4200 | 0.8571 |
| nanobeir / NanoNQ_cosine_recall@3 | 0.5900 | 0.5700 | 0.9661 |
| nanobeir / NanoNQ_cosine_recall@5 | 0.6700 | 0.6500 | 0.9701 |
| nanobeir / NanoNQ_cosine_recall@10 | 0.8000 | 0.7800 | 0.9750 |
| nanobeir / NanoNQ_cosine_ndcg@10 | 0.6371 | 0.6000 | 0.9417 |
| nanobeir / NanoNQ_cosine_mrr@10 | 0.6107 | 0.5613 | 0.9191 |
| nanobeir / NanoNQ_cosine_map@100 | 0.5816 | 0.5433 | 0.9341 |
| nanobeir / NanoQuoraRetrieval_cosine_accuracy@1 | 0.8800 | 0.8800 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_accuracy@3 | 1.0000 | 1.0000 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_accuracy@5 | 1.0000 | 1.0000 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_accuracy@10 | 1.0000 | 1.0000 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_precision@1 | 0.8800 | 0.8800 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_precision@3 | 0.4067 | 0.4067 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_precision@5 | 0.2520 | 0.2520 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_precision@10 | 0.1320 | 0.1320 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_recall@1 | 0.7807 | 0.7807 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_recall@3 | 0.9587 | 0.9587 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_recall@5 | 0.9693 | 0.9693 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_recall@10 | 0.9833 | 0.9833 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_ndcg@10 | 0.9359 | 0.9359 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_mrr@10 | 0.9333 | 0.9333 | 1.0000 |
| nanobeir / NanoQuoraRetrieval_cosine_map@100 | 0.9123 | 0.9123 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_accuracy@1 | 0.4000 | 0.4000 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_accuracy@3 | 0.6400 | 0.6400 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_accuracy@5 | 0.7400 | 0.7400 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_accuracy@10 | 0.8200 | 0.8200 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_precision@1 | 0.4000 | 0.4000 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_precision@3 | 0.3067 | 0.3067 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_precision@5 | 0.2600 | 0.2600 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_precision@10 | 0.1560 | 0.1580 | 1.0128 |
| nanobeir / NanoSCIDOCS_cosine_recall@1 | 0.0847 | 0.0847 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_recall@3 | 0.1897 | 0.1897 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_recall@5 | 0.2667 | 0.2667 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_recall@10 | 0.3187 | 0.3227 | 1.0126 |
| nanobeir / NanoSCIDOCS_cosine_ndcg@10 | 0.3225 | 0.3247 | 1.0068 |
| nanobeir / NanoSCIDOCS_cosine_mrr@10 | 0.5353 | 0.5353 | 1.0000 |
| nanobeir / NanoSCIDOCS_cosine_map@100 | 0.2448 | 0.2454 | 1.0023 |
| nanobeir / NanoArguAna_cosine_accuracy@1 | 0.1000 | 0.0800 | 0.8000 |
| nanobeir / NanoArguAna_cosine_accuracy@3 | 0.4800 | 0.4600 | 0.9583 |
| nanobeir / NanoArguAna_cosine_accuracy@5 | 0.6200 | 0.6400 | 1.0323 |
| nanobeir / NanoArguAna_cosine_accuracy@10 | 0.7200 | 0.7200 | 1.0000 |
| nanobeir / NanoArguAna_cosine_precision@1 | 0.1000 | 0.0800 | 0.8000 |
| nanobeir / NanoArguAna_cosine_precision@3 | 0.1600 | 0.1533 | 0.9583 |
| nanobeir / NanoArguAna_cosine_precision@5 | 0.1240 | 0.1280 | 1.0323 |
| nanobeir / NanoArguAna_cosine_precision@10 | 0.0720 | 0.0720 | 1.0000 |
| nanobeir / NanoArguAna_cosine_recall@1 | 0.1000 | 0.0800 | 0.8000 |
| nanobeir / NanoArguAna_cosine_recall@3 | 0.4800 | 0.4600 | 0.9583 |
| nanobeir / NanoArguAna_cosine_recall@5 | 0.6200 | 0.6400 | 1.0323 |
| nanobeir / NanoArguAna_cosine_recall@10 | 0.7200 | 0.7200 | 1.0000 |
| nanobeir / NanoArguAna_cosine_ndcg@10 | 0.4121 | 0.4062 | 0.9855 |
| nanobeir / NanoArguAna_cosine_mrr@10 | 0.3128 | 0.3046 | 0.9738 |
| nanobeir / NanoArguAna_cosine_map@100 | 0.3267 | 0.3176 | 0.9720 |
| nanobeir / NanoSciFact_cosine_accuracy@1 | 0.6800 | 0.6800 | 1.0000 |
| nanobeir / NanoSciFact_cosine_accuracy@3 | 0.7400 | 0.7400 | 1.0000 |
| nanobeir / NanoSciFact_cosine_accuracy@5 | 0.7400 | 0.7400 | 1.0000 |
| nanobeir / NanoSciFact_cosine_accuracy@10 | 0.7800 | 0.7800 | 1.0000 |
| nanobeir / NanoSciFact_cosine_precision@1 | 0.6800 | 0.6800 | 1.0000 |
| nanobeir / NanoSciFact_cosine_precision@3 | 0.2533 | 0.2533 | 1.0000 |
| nanobeir / NanoSciFact_cosine_precision@5 | 0.1600 | 0.1600 | 1.0000 |
| nanobeir / NanoSciFact_cosine_precision@10 | 0.0880 | 0.0880 | 1.0000 |
| nanobeir / NanoSciFact_cosine_recall@1 | 0.6450 | 0.6450 | 1.0000 |
| nanobeir / NanoSciFact_cosine_recall@3 | 0.7150 | 0.7150 | 1.0000 |
| nanobeir / NanoSciFact_cosine_recall@5 | 0.7250 | 0.7250 | 1.0000 |
| nanobeir / NanoSciFact_cosine_recall@10 | 0.7800 | 0.7800 | 1.0000 |
| nanobeir / NanoSciFact_cosine_ndcg@10 | 0.7209 | 0.7209 | 1.0000 |
| nanobeir / NanoSciFact_cosine_mrr@10 | 0.7117 | 0.7117 | 1.0000 |
| nanobeir / NanoSciFact_cosine_map@100 | 0.7011 | 0.7010 | 0.9999 |
| nanobeir / NanoTouche2020_cosine_accuracy@1 | 0.4898 | 0.4898 | 1.0000 |
| nanobeir / NanoTouche2020_cosine_accuracy@3 | 0.8980 | 0.8980 | 1.0000 |
| nanobeir / NanoTouche2020_cosine_accuracy@5 | 0.9388 | 0.9388 | 1.0000 |
| nanobeir / NanoTouche2020_cosine_accuracy@10 | 0.9796 | 0.9796 | 1.0000 |
| nanobeir / NanoTouche2020_cosine_precision@1 | 0.4898 | 0.4898 | 1.0000 |
| nanobeir / NanoTouche2020_cosine_precision@3 | 0.5442 | 0.5374 | 0.9875 |
| nanobeir / NanoTouche2020_cosine_precision@5 | 0.4816 | 0.4939 | 1.0254 |
| nanobeir / NanoTouche2020_cosine_precision@10 | 0.4000 | 0.4020 | 1.0051 |
| nanobeir / NanoTouche2020_cosine_recall@1 | 0.0309 | 0.0309 | 1.0000 |
| nanobeir / NanoTouche2020_cosine_recall@3 | 0.1093 | 0.1081 | 0.9890 |
| nanobeir / NanoTouche2020_cosine_recall@5 | 0.1638 | 0.1693 | 1.0337 |
| nanobeir / NanoTouche2020_cosine_recall@10 | 0.2602 | 0.2616 | 1.0052 |
| nanobeir / NanoTouche2020_cosine_ndcg@10 | 0.4483 | 0.4509 | 1.0059 |
| nanobeir / NanoTouche2020_cosine_mrr@10 | 0.6885 | 0.6885 | 1.0000 |
| nanobeir / NanoTouche2020_cosine_map@100 | 0.3263 | 0.3282 | 1.0060 |
| nanobeir / NanoBEIR_mean_cosine_accuracy@1 | 0.5054 | 0.4961 | 0.9817 |
| nanobeir / NanoBEIR_mean_cosine_accuracy@3 | 0.6998 | 0.6968 | 0.9956 |
| nanobeir / NanoBEIR_mean_cosine_accuracy@5 | 0.7661 | 0.7661 | 1.0000 |
| nanobeir / NanoBEIR_mean_cosine_accuracy@10 | 0.8354 | 0.8338 | 0.9982 |
| nanobeir / NanoBEIR_mean_cosine_precision@1 | 0.5054 | 0.4961 | 0.9817 |
| nanobeir / NanoBEIR_mean_cosine_precision@3 | 0.3183 | 0.3172 | 0.9967 |
| nanobeir / NanoBEIR_mean_cosine_precision@5 | 0.2494 | 0.2503 | 1.0038 |
| nanobeir / NanoBEIR_mean_cosine_precision@10 | 0.1672 | 0.1675 | 1.0019 |
| nanobeir / NanoBEIR_mean_cosine_recall@1 | 0.3037 | 0.2966 | 0.9766 |
| nanobeir / NanoBEIR_mean_cosine_recall@3 | 0.4591 | 0.4575 | 0.9964 |
| nanobeir / NanoBEIR_mean_cosine_recall@5 | 0.5272 | 0.5277 | 1.0008 |
| nanobeir / NanoBEIR_mean_cosine_recall@10 | 0.5976 | 0.5965 | 0.9982 |
| nanobeir / NanoBEIR_mean_cosine_ndcg@10 | 0.5542 | 0.5514 | 0.9950 |
| nanobeir / NanoBEIR_mean_cosine_mrr@10 | 0.6161 | 0.6111 | 0.9919 |
| nanobeir / NanoBEIR_mean_cosine_map@100 | 0.4769 | 0.4735 | 0.9930 |
Citation
If you use this model or the pruning approach, please cite:
@misc{subedi2025tokenpruning,
author = {Sanjaya Subedi},
title = {Token Embedding Pruning for Sentence Transformers},
year = {2026},
note = {Available at: https://sanjayasubedi.com.np/deeplearning/shrinking-embedding-models-by-pruning-vocabulary/}
}
- Downloads last month
- 17
Model tree for jangedoo/multilingual-e5-small-en-pruned
Base model
intfloat/multilingual-e5-small