Multilingual base SWU Matryoshka
This is a sentence-transformers model finetuned from intfloat/multilingual-e5-small on the json dataset. 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 tokens
- Similarity Function: Cosine Similarity
- Training Dataset:
- json
- Language: multilingual
- License: apache-2.0
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("ValentinaKim/Multilingual-base-SWU-Matryoshka")
# Run inference
sentences = [
'학교생활기록부 교과성적 대체 점수(비교내신) 대상자는 논술(논술우수자전형), 실기/실적(실기우수자전형_체육) 지원자 중 고등학교 졸업학력 검정고시 출신 지원자 및 교과성적 산출 불가자입니다.',
'학교생활기록부 교과성적 대체 점수(비교내신) 대상자를 알려줘.',
'청소년학전공의 교육 내용은 무엇입니까?',
]
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]
Evaluation
Metrics
Information Retrieval
- Dataset:
dim_256
- Evaluated with
InformationRetrievalEvaluator
Metric | Value |
---|---|
cosine_accuracy@1 | 0.6364 |
cosine_accuracy@3 | 0.9091 |
cosine_accuracy@5 | 1.0 |
cosine_accuracy@10 | 1.0 |
cosine_precision@1 | 0.6364 |
cosine_precision@3 | 0.303 |
cosine_precision@5 | 0.2 |
cosine_precision@10 | 0.1 |
cosine_recall@1 | 0.6364 |
cosine_recall@3 | 0.9091 |
cosine_recall@5 | 1.0 |
cosine_recall@10 | 1.0 |
cosine_ndcg@10 | 0.8476 |
cosine_mrr@10 | 0.7955 |
cosine_map@100 | 0.7955 |
Information Retrieval
- Dataset:
dim_128
- Evaluated with
InformationRetrievalEvaluator
Metric | Value |
---|---|
cosine_accuracy@1 | 0.6364 |
cosine_accuracy@3 | 0.9091 |
cosine_accuracy@5 | 1.0 |
cosine_accuracy@10 | 1.0 |
cosine_precision@1 | 0.6364 |
cosine_precision@3 | 0.303 |
cosine_precision@5 | 0.2 |
cosine_precision@10 | 0.1 |
cosine_recall@1 | 0.6364 |
cosine_recall@3 | 0.9091 |
cosine_recall@5 | 1.0 |
cosine_recall@10 | 1.0 |
cosine_ndcg@10 | 0.8476 |
cosine_mrr@10 | 0.7955 |
cosine_map@100 | 0.7955 |
Information Retrieval
- Dataset:
dim_64
- Evaluated with
InformationRetrievalEvaluator
Metric | Value |
---|---|
cosine_accuracy@1 | 0.6364 |
cosine_accuracy@3 | 0.9091 |
cosine_accuracy@5 | 1.0 |
cosine_accuracy@10 | 1.0 |
cosine_precision@1 | 0.6364 |
cosine_precision@3 | 0.303 |
cosine_precision@5 | 0.2 |
cosine_precision@10 | 0.1 |
cosine_recall@1 | 0.6364 |
cosine_recall@3 | 0.9091 |
cosine_recall@5 | 1.0 |
cosine_recall@10 | 1.0 |
cosine_ndcg@10 | 0.8357 |
cosine_mrr@10 | 0.7803 |
cosine_map@100 | 0.7803 |
Training Details
Training Dataset
json
- Dataset: json
- Size: 94 training samples
- Columns:
positive
andanchor
- Approximate statistics based on the first 94 samples:
positive anchor type string string details - min: 24 tokens
- mean: 89.93 tokens
- max: 272 tokens
- min: 10 tokens
- mean: 19.18 tokens
- max: 35 tokens
- Samples:
positive anchor 서울여자대학교 수시모집에서 평가하는 요소는 다음과 같습니다. 1. 서류 평가(학업역량 40%, 진로역량 35%, 공동체역량 25%) 2. 면접 평가(인성 및 의사소통능력, 발전가능성) 3. 학교생활기록부에 학교폭력 관련 기재사항이 있을 경우, 정성평가로 반영합니다.
서울여자대학교 수시모집에서 평가하는 요소를 알려줘.
서울여자대학교 학생부종합전형 지원자에게 필요한 지원자격 정보는 다음과 같습니다. 지원자격은 기초생활수급자, 차상위계층, 한부모가족 지원대상자, 국가보훈대상자, 자립지원 대상 아동, 농어촌학생 등입니다. 각 지원자격에 따라 필요한 제출서류가 다르므로, 지원자격에 따라 필요한 제출서류를 확인하여야 합니다.
서울여자대학교 학생부종합전형 지원자에게 필요한 지원자격 정보를 알려줘.
SWU의 SI(Social Innovation)교육은 사회적 가치 확산을 위해 혁신적인 방법론을 적용하여 긍정적인 사회 변화를 유도하는 서울여자대학교만의 차별화된 교육입니다. 바롬종합설계프로젝트는 유네스코한국위원회가 인증한 유네스코지속가능발전교육공식프로젝트입니다.
SWU의 SI(Social Innovation)교육에 대해 알려줘.
- Loss:
MatryoshkaLoss
with these parameters:{ "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1 ], "n_dims_per_step": -1 }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy
: epochgradient_accumulation_steps
: 16learning_rate
: 2e-05num_train_epochs
: 4lr_scheduler_type
: cosinewarmup_ratio
: 0.1tf32
: Falseload_best_model_at_end
: Trueoptim
: adamw_torch_fusedbatch_sampler
: no_duplicates
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: epochprediction_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
: 16eval_accumulation_steps
: Nonelearning_rate
: 2e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1.0num_train_epochs
: 4max_steps
: -1lr_scheduler_type
: cosinelr_scheduler_kwargs
: {}warmup_ratio
: 0.1warmup_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
: Falselocal_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
: Trueignore_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_torch_fusedoptim_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
: Falseeval_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
: Falsebatch_sampler
: no_duplicatesmulti_dataset_batch_sampler
: proportional
Training Logs
Epoch | Step | dim_128_cosine_map@100 | dim_256_cosine_map@100 | dim_64_cosine_map@100 |
---|---|---|---|---|
1.0 | 1 | 0.7955 | 0.7955 | 0.7803 |
2.0 | 2 | 0.7955 | 0.7955 | 0.7803 |
3.0 | 4 | 0.7955 | 0.7955 | 0.7803 |
1.0 | 1 | 0.7955 | 0.7955 | 0.7803 |
2.0 | 2 | 0.7955 | 0.7955 | 0.7803 |
3.0 | 4 | 0.7955 | 0.7955 | 0.7803 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.10.14
- Sentence Transformers: 3.1.1
- Transformers: 4.41.2
- PyTorch: 2.1.2+cu121
- Accelerate: 0.34.2
- Datasets: 2.19.1
- Tokenizers: 0.19.1
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",
}
MatryoshkaLoss
@misc{kusupati2024matryoshka,
title={Matryoshka Representation Learning},
author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
year={2024},
eprint={2205.13147},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
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
- 14
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.
Model tree for ValentinaKim/Multilingual-base-SWU-Matryoshka
Base model
intfloat/multilingual-e5-smallEvaluation results
- Cosine Accuracy@1 on dim 256self-reported0.636
- Cosine Accuracy@3 on dim 256self-reported0.909
- Cosine Accuracy@5 on dim 256self-reported1.000
- Cosine Accuracy@10 on dim 256self-reported1.000
- Cosine Precision@1 on dim 256self-reported0.636
- Cosine Precision@3 on dim 256self-reported0.303
- Cosine Precision@5 on dim 256self-reported0.200
- Cosine Precision@10 on dim 256self-reported0.100
- Cosine Recall@1 on dim 256self-reported0.636
- Cosine Recall@3 on dim 256self-reported0.909