SentenceTransformer based on intfloat/multilingual-e5-small
This is a sentence-transformers model finetuned from intfloat/multilingual-e5-small. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
Model Details
Model Description
- Model Type: Sentence Transformer
- Base model: intfloat/multilingual-e5-small
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 384 dimensions
- Similarity Function: Cosine Similarity
Model Sources
- Documentation: Sentence Transformers Documentation
- Repository: Sentence Transformers on GitHub
- Hugging Face: Sentence Transformers on Hugging Face
Full Model Architecture
SentenceTransformer(
(0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel
(1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
(2): Normalize()
)
Usage
Direct Usage (Sentence Transformers)
First install the Sentence Transformers library:
pip install -U sentence-transformers
Then you can load this model and run inference.
from sentence_transformers import SentenceTransformer
# Download from the 🤗 Hub
model = SentenceTransformer("agentlans/multilingual-e5-small-aligned")
# Run inference
sentences = [
'What do they think it is that prevents the products of human ingenuity from being themselves, fruits of the tree of life, and hence, in some sense, obeying evolutionary rules?',
'Կարծում եք ի՞նչն է խանգարում, որ մարդկային հնարամտության արդյունքները իրենք էլ լինեն կյանքի ծառի պտուղներ և այդպիսով ինչ-որ իմաստով ենթարկվեն էվոլուցիայի կանոններին:',
'(Smiech) No dobre, idem do Ameriky.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
Training Details
Training Dataset
Unnamed Dataset
- Size: 867,042 training samples
- Columns:
sentence_0
andsentence_1
- Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1 type string string details - min: 3 tokens
- mean: 21.83 tokens
- max: 177 tokens
- min: 4 tokens
- mean: 24.92 tokens
- max: 229 tokens
- Samples:
sentence_0 sentence_1 I like English best of all subjects.
Tykkään englannista eniten kaikista aineista.
We shall offer negotiations. Quite right.
- Oferecer-nos-emos para negociar.
It was soon learned that Zelaya had been taken to Costa Rica, where he continued to call himself as the legal head of state.
Al snel werd bekend dat Zelaya naar Costa Rica was overgebracht, waar hij zich nog steeds het officiële staatshoofd noemde.
- Loss:
MultipleNegativesRankingLoss
with these parameters:{ "scale": 20.0, "similarity_fct": "cos_sim" }
Training Hyperparameters
Non-Default Hyperparameters
num_train_epochs
: 1multi_dataset_batch_sampler
: round_robin
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: noprediction_loss_only
: Trueper_device_train_batch_size
: 8per_device_eval_batch_size
: 8per_gpu_train_batch_size
: Noneper_gpu_eval_batch_size
: Nonegradient_accumulation_steps
: 1eval_accumulation_steps
: Nonetorch_empty_cache_steps
: Nonelearning_rate
: 5e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1num_train_epochs
: 1max_steps
: -1lr_scheduler_type
: linearlr_scheduler_kwargs
: {}warmup_ratio
: 0.0warmup_steps
: 0log_level
: passivelog_level_replica
: warninglog_on_each_node
: Truelogging_nan_inf_filter
: Truesave_safetensors
: Truesave_on_each_node
: Falsesave_only_model
: Falserestore_callback_states_from_checkpoint
: Falseno_cuda
: Falseuse_cpu
: Falseuse_mps_device
: Falseseed
: 42data_seed
: Nonejit_mode_eval
: Falseuse_ipex
: Falsebf16
: Falsefp16
: Falsefp16_opt_level
: O1half_precision_backend
: autobf16_full_eval
: Falsefp16_full_eval
: Falsetf32
: Nonelocal_rank
: 0ddp_backend
: Nonetpu_num_cores
: Nonetpu_metrics_debug
: Falsedebug
: []dataloader_drop_last
: Falsedataloader_num_workers
: 0dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Falseignore_data_skip
: Falsefsdp
: []fsdp_min_num_params
: 0fsdp_config
: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap
: Noneaccelerator_config
: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}deepspeed
: Nonelabel_smoothing_factor
: 0.0optim
: adamw_torchoptim_args
: Noneadafactor
: Falsegroup_by_length
: Falselength_column_name
: lengthddp_find_unused_parameters
: Noneddp_bucket_cap_mb
: Noneddp_broadcast_buffers
: Falsedataloader_pin_memory
: Truedataloader_persistent_workers
: Falseskip_memory_metrics
: Trueuse_legacy_prediction_loop
: Falsepush_to_hub
: Falseresume_from_checkpoint
: Nonehub_model_id
: Nonehub_strategy
: every_savehub_private_repo
: Falsehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseinclude_for_metrics
: []eval_do_concat_batches
: Truefp16_backend
: autopush_to_hub_model_id
: Nonepush_to_hub_organization
: Nonemp_parameters
:auto_find_batch_size
: Falsefull_determinism
: Falsetorchdynamo
: Noneray_scope
: lastddp_timeout
: 1800torch_compile
: Falsetorch_compile_backend
: Nonetorch_compile_mode
: Nonedispatch_batches
: Nonesplit_batches
: Noneinclude_tokens_per_second
: Falseinclude_num_input_tokens_seen
: Falseneftune_noise_alpha
: Noneoptim_target_modules
: Nonebatch_eval_metrics
: Falseeval_on_start
: Falseuse_liger_kernel
: Falseeval_use_gather_object
: Falseaverage_tokens_across_devices
: Falseprompts
: Nonebatch_sampler
: batch_samplermulti_dataset_batch_sampler
: round_robin
Training Logs
Click to expand
Epoch | Step | Training Loss |
---|---|---|
0.0046 | 500 | 0.0378 |
0.0092 | 1000 | 0.0047 |
0.0138 | 1500 | 0.006 |
0.0185 | 2000 | 0.0045 |
0.0231 | 2500 | 0.0027 |
0.0277 | 3000 | 0.005 |
0.0323 | 3500 | 0.0045 |
0.0369 | 4000 | 0.005 |
0.0415 | 4500 | 0.0066 |
0.0461 | 5000 | 0.0029 |
0.0507 | 5500 | 0.0041 |
0.0554 | 6000 | 0.0064 |
0.0600 | 6500 | 0.0044 |
0.0646 | 7000 | 0.0039 |
0.0692 | 7500 | 0.0025 |
0.0738 | 8000 | 0.0026 |
0.0784 | 8500 | 0.0036 |
0.0830 | 9000 | 0.0027 |
0.0877 | 9500 | 0.0015 |
0.0923 | 10000 | 0.003 |
0.0969 | 10500 | 0.0013 |
0.1015 | 11000 | 0.002 |
0.1061 | 11500 | 0.0038 |
0.1107 | 12000 | 0.0017 |
0.1153 | 12500 | 0.0029 |
0.1199 | 13000 | 0.0032 |
0.1246 | 13500 | 0.0036 |
0.1292 | 14000 | 0.004 |
0.1338 | 14500 | 0.0036 |
0.1384 | 15000 | 0.0025 |
0.1430 | 15500 | 0.0022 |
0.1476 | 16000 | 0.0017 |
0.1522 | 16500 | 0.0019 |
0.1569 | 17000 | 0.0022 |
0.1615 | 17500 | 0.0028 |
0.1661 | 18000 | 0.0033 |
0.1707 | 18500 | 0.0025 |
0.1753 | 19000 | 0.0014 |
0.1799 | 19500 | 0.0033 |
0.1845 | 20000 | 0.0023 |
0.1891 | 20500 | 0.0023 |
0.1938 | 21000 | 0.0009 |
0.1984 | 21500 | 0.0043 |
0.2030 | 22000 | 0.0021 |
0.2076 | 22500 | 0.0025 |
0.2122 | 23000 | 0.0017 |
0.2168 | 23500 | 0.0024 |
0.2214 | 24000 | 0.0021 |
0.2261 | 24500 | 0.0023 |
0.2307 | 25000 | 0.0014 |
0.2353 | 25500 | 0.0027 |
0.2399 | 26000 | 0.0025 |
0.2445 | 26500 | 0.0022 |
0.2491 | 27000 | 0.0022 |
0.2537 | 27500 | 0.0024 |
0.2583 | 28000 | 0.0035 |
0.2630 | 28500 | 0.0032 |
0.2676 | 29000 | 0.0048 |
0.2722 | 29500 | 0.0008 |
0.2768 | 30000 | 0.0027 |
0.2814 | 30500 | 0.004 |
0.2860 | 31000 | 0.0013 |
0.2906 | 31500 | 0.002 |
0.2953 | 32000 | 0.0016 |
0.2999 | 32500 | 0.0027 |
0.3045 | 33000 | 0.0014 |
0.3091 | 33500 | 0.0022 |
0.3137 | 34000 | 0.0017 |
0.3183 | 34500 | 0.0022 |
0.3229 | 35000 | 0.0026 |
0.3275 | 35500 | 0.003 |
0.3322 | 36000 | 0.0022 |
0.3368 | 36500 | 0.0022 |
0.3414 | 37000 | 0.0018 |
0.3460 | 37500 | 0.0028 |
0.3506 | 38000 | 0.0018 |
0.3552 | 38500 | 0.0037 |
0.3598 | 39000 | 0.003 |
0.3645 | 39500 | 0.002 |
0.3691 | 40000 | 0.001 |
0.3737 | 40500 | 0.0015 |
0.3783 | 41000 | 0.0023 |
0.3829 | 41500 | 0.0017 |
0.3875 | 42000 | 0.0034 |
0.3921 | 42500 | 0.0016 |
0.3967 | 43000 | 0.0019 |
0.4014 | 43500 | 0.0015 |
0.4060 | 44000 | 0.0026 |
0.4106 | 44500 | 0.0012 |
0.4152 | 45000 | 0.0014 |
0.4198 | 45500 | 0.0027 |
0.4244 | 46000 | 0.0016 |
0.4290 | 46500 | 0.0027 |
0.4337 | 47000 | 0.0033 |
0.4383 | 47500 | 0.0023 |
0.4429 | 48000 | 0.0024 |
0.4475 | 48500 | 0.0019 |
0.4521 | 49000 | 0.0017 |
0.4567 | 49500 | 0.004 |
0.4613 | 50000 | 0.0036 |
0.4659 | 50500 | 0.001 |
0.4706 | 51000 | 0.0016 |
0.4752 | 51500 | 0.0024 |
0.4798 | 52000 | 0.0009 |
0.4844 | 52500 | 0.0011 |
0.4890 | 53000 | 0.0018 |
0.4936 | 53500 | 0.0012 |
0.4982 | 54000 | 0.0012 |
0.5029 | 54500 | 0.0014 |
0.5075 | 55000 | 0.0025 |
0.5121 | 55500 | 0.0016 |
0.5167 | 56000 | 0.0015 |
0.5213 | 56500 | 0.002 |
0.5259 | 57000 | 0.0008 |
0.5305 | 57500 | 0.0017 |
0.5351 | 58000 | 0.0015 |
0.5398 | 58500 | 0.0009 |
0.5444 | 59000 | 0.0019 |
0.5490 | 59500 | 0.0014 |
0.5536 | 60000 | 0.0028 |
0.5582 | 60500 | 0.0014 |
0.5628 | 61000 | 0.0032 |
0.5674 | 61500 | 0.0013 |
0.5721 | 62000 | 0.002 |
0.5767 | 62500 | 0.0018 |
0.5813 | 63000 | 0.0015 |
0.5859 | 63500 | 0.0008 |
0.5905 | 64000 | 0.0021 |
0.5951 | 64500 | 0.0008 |
0.5997 | 65000 | 0.002 |
0.6043 | 65500 | 0.0023 |
0.6090 | 66000 | 0.0022 |
0.6136 | 66500 | 0.0013 |
0.6182 | 67000 | 0.0011 |
0.6228 | 67500 | 0.0014 |
0.6274 | 68000 | 0.0027 |
0.6320 | 68500 | 0.002 |
0.6366 | 69000 | 0.0013 |
0.6413 | 69500 | 0.0026 |
0.6459 | 70000 | 0.0014 |
0.6505 | 70500 | 0.0017 |
0.6551 | 71000 | 0.0023 |
0.6597 | 71500 | 0.0025 |
0.6643 | 72000 | 0.0013 |
0.6689 | 72500 | 0.0008 |
0.6735 | 73000 | 0.0017 |
0.6782 | 73500 | 0.0022 |
0.6828 | 74000 | 0.0021 |
0.6874 | 74500 | 0.0008 |
0.6920 | 75000 | 0.0007 |
0.6966 | 75500 | 0.0038 |
0.7012 | 76000 | 0.0011 |
0.7058 | 76500 | 0.0016 |
0.7105 | 77000 | 0.0013 |
0.7151 | 77500 | 0.0042 |
0.7197 | 78000 | 0.0009 |
0.7243 | 78500 | 0.0004 |
0.7289 | 79000 | 0.0006 |
0.7335 | 79500 | 0.0007 |
0.7381 | 80000 | 0.0014 |
0.7428 | 80500 | 0.002 |
0.7474 | 81000 | 0.0017 |
0.7520 | 81500 | 0.0014 |
0.7566 | 82000 | 0.0015 |
0.7612 | 82500 | 0.0013 |
0.7658 | 83000 | 0.001 |
0.7704 | 83500 | 0.0019 |
0.7750 | 84000 | 0.0009 |
0.7797 | 84500 | 0.0021 |
0.7843 | 85000 | 0.0015 |
0.7889 | 85500 | 0.001 |
0.7935 | 86000 | 0.0008 |
0.7981 | 86500 | 0.0039 |
0.8027 | 87000 | 0.0018 |
0.8073 | 87500 | 0.0009 |
0.8120 | 88000 | 0.0018 |
0.8166 | 88500 | 0.0008 |
0.8212 | 89000 | 0.0007 |
0.8258 | 89500 | 0.0009 |
0.8304 | 90000 | 0.002 |
0.8350 | 90500 | 0.001 |
0.8396 | 91000 | 0.0007 |
0.8442 | 91500 | 0.0008 |
0.8489 | 92000 | 0.0021 |
0.8535 | 92500 | 0.0013 |
0.8581 | 93000 | 0.0009 |
0.8627 | 93500 | 0.002 |
0.8673 | 94000 | 0.0012 |
0.8719 | 94500 | 0.0034 |
0.8765 | 95000 | 0.0027 |
0.8812 | 95500 | 0.0006 |
0.8858 | 96000 | 0.002 |
0.8904 | 96500 | 0.0005 |
0.8950 | 97000 | 0.0009 |
0.8996 | 97500 | 0.0007 |
0.9042 | 98000 | 0.0015 |
0.9088 | 98500 | 0.0006 |
0.9134 | 99000 | 0.0004 |
0.9181 | 99500 | 0.0006 |
0.9227 | 100000 | 0.0031 |
0.9273 | 100500 | 0.0013 |
0.9319 | 101000 | 0.0024 |
0.9365 | 101500 | 0.0006 |
0.9411 | 102000 | 0.0017 |
0.9457 | 102500 | 0.0007 |
0.9504 | 103000 | 0.0012 |
0.9550 | 103500 | 0.0011 |
0.9596 | 104000 | 0.0007 |
0.9642 | 104500 | 0.0004 |
0.9688 | 105000 | 0.0021 |
0.9734 | 105500 | 0.0027 |
0.9780 | 106000 | 0.0016 |
0.9826 | 106500 | 0.0022 |
0.9873 | 107000 | 0.0017 |
0.9919 | 107500 | 0.0009 |
0.9965 | 108000 | 0.0008 |
Framework Versions
- Python: 3.10.12
- Sentence Transformers: 3.3.0
- Transformers: 4.46.3
- PyTorch: 2.5.1+cu124
- Accelerate: 1.1.1
- Datasets: 3.1.0
- Tokenizers: 0.20.3
Citation
BibTeX
Sentence Transformers
@inproceedings{reimers-2019-sentence-bert,
title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
author = "Reimers, Nils and Gurevych, Iryna",
booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
month = "11",
year = "2019",
publisher = "Association for Computational Linguistics",
url = "https://arxiv.org/abs/1908.10084",
}
MultipleNegativesRankingLoss
@misc{henderson2017efficient,
title={Efficient Natural Language Response Suggestion for Smart Reply},
author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
year={2017},
eprint={1705.00652},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
- Downloads last month
- 41
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.