Add new SentenceTransformer model

782dadd verified 8 months ago

50.9 kB

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:14495
  - loss:MultipleNegativesRankingLoss
base_model: intfloat/multilingual-e5-base
widget:
  - source_sentence: tôi muốn mua máy rửa chén dung tích trên 12 bộ và giá nhỏ hơn 16 triệu
    sentences:
      - >-
        Máy rửa chén Bosch SMS25CI05E, Dung tích 13 bộ, Tiết kiệm năng lượng A+,
        Giá: 13.900.000
      - 'Bàn học gỗ MDF, Rộng 1m2, Ngăn kéo bên, Giá: 2.400.000'
      - >-
        Philips HR2041/10 - . Công suất 450W, cối thủy tinh 1.25 lít, 2 tốc độ,
        lưỡi dao inox chống gỉ, dễ tháo rửa. Kích thước: 210 x 180 x 370 mm.
        Trọng lượng: 2.2 kg. Giá: 990.000 VNĐ
  - source_sentence: >-
      mình muốn tủ đông đứng dung tích 220 lít, công suất 200W, giá từ 5 đến 6
      triệu
    sentences:
      - 'Tủ đông đứng Sanaky 220L, công suất 200W, Giá: 5.950.000'
      - 'Đàn piano điện Yamaha P-125, 88 phím, Trọng lượng 12kg, Giá: 15.900.000'
      - 'Sofa da thật 3 chỗ, Nặng 55kg, Khung gỗ sồi, Giá: 18.500.000'
  - source_sentence: tôi muốn mua ghế văn phòng chịu tải tối thiểu 138kg
    sentences:
      - >-
        Ghế công thái học Sihoo M18, Tải trọng 150kg, Tựa lưng lưới, Giá:
        3.800.000
      - 'Ghế xoay lưới Ergonomic, Khung thép, Tải trọng 150kg, Giá: 3.800.000'
      - >-
        Quạt điều hòa Sunhouse SHD7725 - . Bình nước 45 lít, công suất 220W, lưu
        lượng 5.000 m³/h, điều khiển từ xa. Trọng lượng: 16 kg. Giá: 4.490.000
        VNĐ
  - source_sentence: Có máy sưởi điện công suất không quá 2240W không
    sentences:
      - >-
        Smart TV Samsung 43AU7700, Màn hình 43 inch, 4K UHD, Hỗ trợ HDR, Giá:
        8.900.000
      - 'Loa Soundbar Samsung Q990B, Công suất 440W, Dolby Atmos, Giá: 21.500.000'
      - >-
        Máy sưởi dầu FujiE OFR4413 - . Công suất 2000W, 13 thanh sưởi, có bánh
        xe di chuyển. Giá: 3.490.000 VNĐ
  - source_sentence: có đồng hồ cơ khả năng trữ cót tối thiểu 26 giờ
    sentences:
      - 'Vali Trip P803, size 24 inch, chất liệu ABS, khóa TSA, Giá: 1.750.000'
      - 'Đèn bàn LED Xiaomi Mi, công suất 10W, điều chỉnh độ sáng, Giá: 470.000'
      - 'Đồng hồ Seiko 5 SNK809, Automatic, Trữ cót 40 giờ, Giá: 4.200.000'
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@2
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_accuracy@100
  - cosine_precision@1
  - cosine_precision@2
  - cosine_precision@5
  - cosine_precision@10
  - cosine_precision@100
  - cosine_recall@1
  - cosine_recall@2
  - cosine_recall@5
  - cosine_recall@10
  - cosine_recall@100
  - cosine_ndcg@10
  - cosine_mrr@1
  - cosine_mrr@2
  - cosine_mrr@5
  - cosine_mrr@10
  - cosine_mrr@100
  - cosine_map@100
model-index:
  - name: SentenceTransformer based on intfloat/multilingual-e5-base
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: cosine_accuracy@1
            value: 0.281812538795779
            name: Cosine Accuracy@1
          - type: cosine_accuracy@2
            value: 0.44009931719428924
            name: Cosine Accuracy@2
          - type: cosine_accuracy@5
            value: 0.675356921166977
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.8392302917442582
            name: Cosine Accuracy@10
          - type: cosine_accuracy@100
            value: 1
            name: Cosine Accuracy@100
          - type: cosine_precision@1
            value: 0.281812538795779
            name: Cosine Precision@1
          - type: cosine_precision@2
            value: 0.22004965859714462
            name: Cosine Precision@2
          - type: cosine_precision@5
            value: 0.13507138423339543
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.08392302917442582
            name: Cosine Precision@10
          - type: cosine_precision@100
            value: 0.01
            name: Cosine Precision@100
          - type: cosine_recall@1
            value: 0.281812538795779
            name: Cosine Recall@1
          - type: cosine_recall@2
            value: 0.44009931719428924
            name: Cosine Recall@2
          - type: cosine_recall@5
            value: 0.675356921166977
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.8392302917442582
            name: Cosine Recall@10
          - type: cosine_recall@100
            value: 1
            name: Cosine Recall@100
          - type: cosine_ndcg@10
            value: 0.540303991990625
            name: Cosine Ndcg@10
          - type: cosine_mrr@1
            value: 0.281812538795779
            name: Cosine Mrr@1
          - type: cosine_mrr@2
            value: 0.3609559279950341
            name: Cosine Mrr@2
          - type: cosine_mrr@5
            value: 0.4252017380509011
            name: Cosine Mrr@5
          - type: cosine_mrr@10
            value: 0.446985752712011
            name: Cosine Mrr@10
          - type: cosine_mrr@100
            value: 0.45749681159297556
            name: Cosine Mrr@100
          - type: cosine_map@100
            value: 0.45749681159297456
            name: Cosine Map@100

SentenceTransformer based on intfloat/multilingual-e5-base

This is a sentence-transformers model finetuned from intfloat/multilingual-e5-base. It maps sentences & paragraphs to a 768-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-base
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 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: XLMRobertaModel 
  (1): Pooling({'word_embedding_dimension': 768, '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("meandyou200175/sp_chatbot_query_e5")
# Run inference
sentences = [
    'có đồng hồ cơ khả năng trữ cót tối thiểu 26 giờ',
    'Đồng hồ Seiko 5 SNK809, Automatic, Trữ cót 40 giờ, Giá: 4.200.000',
    'Vali Trip P803, size 24 inch, chất liệu ABS, khóa TSA, Giá: 1.750.000',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Information Retrieval

Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.2818
cosine_accuracy@2	0.4401
cosine_accuracy@5	0.6754
cosine_accuracy@10	0.8392
cosine_accuracy@100	1.0
cosine_precision@1	0.2818
cosine_precision@2	0.22
cosine_precision@5	0.1351
cosine_precision@10	0.0839
cosine_precision@100	0.01
cosine_recall@1	0.2818
cosine_recall@2	0.4401
cosine_recall@5	0.6754
cosine_recall@10	0.8392
cosine_recall@100	1.0
cosine_ndcg@10	0.5403
cosine_mrr@1	0.2818
cosine_mrr@2	0.361
cosine_mrr@5	0.4252
cosine_mrr@10	0.447
cosine_mrr@100	0.4575
cosine_map@100	0.4575

Training Details

Training Dataset

Unnamed Dataset

Size: 14,495 training samples
Columns: query and positive
Approximate statistics based on the first 1000 samples:
query positive
type string string
details
min: 5 tokens
mean: 14.96 tokens
max: 28 tokens

min: 19 tokens
mean: 37.14 tokens
max: 137 tokens

	query	positive
type	string	string
details	min: 5 tokens mean: 14.96 tokens max: 28 tokens	min: 19 tokens mean: 37.14 tokens max: 137 tokens

Samples:

query	positive
`Có máy làm mát không khí dưới 3.043.000.000 VNĐ không`	`Sunhouse SHD7713 - . Công suất 80W, dung tích 8 lít, 3 tốc độ gió, chế độ hẹn giờ, quạt gió hơi nước, vận hành êm, dễ di chuyển, thân thiện môi trường. Kích thước: 400 x 300 x 700 mm. Trọng lượng: 5 kg. Giá: 1.790.000 VNĐ`
`mình cần robot hút bụi pin lớn hơn 3380mAh`	`Robot hút bụi Dreame L10 Pro, Pin 5200mAh, Lực hút 4000Pa, Giá: 9.200.000`
`Bạn có ghế massage Daikiosan công suất lớn hơn 126W không ạ?`	`Ghế massage Daikiosan DK-200, công suất 140W, nhiều chế độ, Giá: 4.250.000`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Evaluation Dataset

Unnamed Dataset

Size: 1,611 evaluation samples
Columns: query and positive
Approximate statistics based on the first 1000 samples:
query positive
type string string
details
min: 7 tokens
mean: 15.16 tokens
max: 28 tokens

min: 18 tokens
mean: 38.32 tokens
max: 140 tokens

	query	positive
type	string	string
details	min: 7 tokens mean: 15.16 tokens max: 28 tokens	min: 18 tokens mean: 38.32 tokens max: 140 tokens

Samples:

query	positive
`có công tắc thông minh giá dưới 660.000 không`	`Công tắc Xiaomi Smart WiFi, chịu tải 16A, kết nối 2.4GHz, Giá: 420.000`
`Có robot hút bụi công suất hút trên 2000Pa và pin ít nhất 3536mAh không`	`Robot hút bụi Roborock S6 MaxV - . Công suất hút 2500Pa, pin 5.200 mAh, chạy 180 phút, điều hướng Lidar. Trọng lượng: 3.7 kg. Giá: 12.900.000 VNĐ`
`mình cần bếp từ công suất tối thiểu 1932W`	`Bếp từ Midea MI-T2117DC, Công suất 2100W, 8 chế độ nấu, Giá: 1.750.000`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 2
per_device_eval_batch_size: 2
learning_rate: 2e-05
num_train_epochs: 6
warmup_ratio: 0.1
fp16: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 2
per_device_eval_batch_size: 2
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 2e-05
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 6
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.1
warmup_steps: 0
log_level: passive
log_level_replica: warning
log_on_each_node: True
logging_nan_inf_filter: True
save_safetensors: True
save_on_each_node: False
save_only_model: False
restore_callback_states_from_checkpoint: False
no_cuda: False
use_cpu: False
use_mps_device: False
seed: 42
data_seed: None
jit_mode_eval: False
use_ipex: False
bf16: False
fp16: True
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: 0
ddp_backend: None
tpu_num_cores: None
tpu_metrics_debug: False
debug: []
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_prefetch_factor: None
past_index: -1
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: False
ignore_data_skip: False
fsdp: []
fsdp_min_num_params: 0
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
fsdp_transformer_layer_cls_to_wrap: None
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch
optim_args: None
adafactor: False
group_by_length: False
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
dataloader_pin_memory: True
dataloader_persistent_workers: False
skip_memory_metrics: True
use_legacy_prediction_loop: False
push_to_hub: False
resume_from_checkpoint: None
hub_model_id: None
hub_strategy: every_save
hub_private_repo: None
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
include_for_metrics: []
eval_do_concat_batches: True
fp16_backend: auto
push_to_hub_model_id: None
push_to_hub_organization: None
mp_parameters:
auto_find_batch_size: False
full_determinism: False
torchdynamo: None
ray_scope: last
ddp_timeout: 1800
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
include_tokens_per_second: False
include_num_input_tokens_seen: False
neftune_noise_alpha: None
optim_target_modules: None
batch_eval_metrics: False
eval_on_start: False
use_liger_kernel: False
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	cosine_ndcg@10
-1	-1	-	-	0.3399
0.0138	100	0.3618	-	-
0.0276	200	0.2506	-	-
0.0414	300	0.1367	-	-
0.0552	400	0.0286	-	-
0.0690	500	0.0263	-	-
0.0828	600	0.0381	-	-
0.0966	700	0.0129	-	-
0.1104	800	0.0138	-	-
0.1242	900	0.0125	-	-
0.1380	1000	0.0175	0.0200	0.3713
0.1518	1100	0.0146	-	-
0.1656	1200	0.0293	-	-
0.1794	1300	0.014	-	-
0.1932	1400	0.0137	-	-
0.2070	1500	0.0082	-	-
0.2208	1600	0.0134	-	-
0.2345	1700	0.0128	-	-
0.2483	1800	0.0032	-	-
0.2621	1900	0.0129	-	-
0.2759	2000	0.0141	0.0188	0.3510
0.2897	2100	0.058	-	-
0.3035	2200	0.02	-	-
0.3173	2300	0.0273	-	-
0.3311	2400	0.0269	-	-
0.3449	2500	0.0064	-	-
0.3587	2600	0.0078	-	-
0.3725	2700	0.0383	-	-
0.3863	2800	0.0017	-	-
0.4001	2900	0.0274	-	-
0.4139	3000	0.0304	0.0104	0.4526
0.4277	3100	0.0018	-	-
0.4415	3200	0.0082	-	-
0.4553	3300	0.0177	-	-
0.4691	3400	0.0117	-	-
0.4829	3500	0.0135	-	-
0.4967	3600	0.0362	-	-
0.5105	3700	0.0067	-	-
0.5243	3800	0.0009	-	-
0.5381	3900	0.0139	-	-
0.5519	4000	0.0046	0.0099	0.4424
0.5657	4100	0.0037	-	-
0.5795	4200	0.011	-	-
0.5933	4300	0.0187	-	-
0.6071	4400	0.0244	-	-
0.6209	4500	0.0032	-	-
0.6347	4600	0.0086	-	-
0.6485	4700	0.0398	-	-
0.6623	4800	0.0187	-	-
0.6760	4900	0.0012	-	-
0.6898	5000	0.0095	0.0170	0.4107
0.7036	5100	0.0183	-	-
0.7174	5200	0.0386	-	-
0.7312	5300	0.0072	-	-
0.7450	5400	0.0118	-	-
0.7588	5500	0.0035	-	-
0.7726	5600	0.0103	-	-
0.7864	5700	0.0093	-	-
0.8002	5800	0.0237	-	-
0.8140	5900	0.0079	-	-
0.8278	6000	0.0096	0.0116	0.4449
0.8416	6100	0.014	-	-
0.8554	6200	0.0092	-	-
0.8692	6300	0.0227	-	-
0.8830	6400	0.0022	-	-
0.8968	6500	0.0097	-	-
0.9106	6600	0.0136	-	-
0.9244	6700	0.0122	-	-
0.9382	6800	0.0177	-	-
0.9520	6900	0.0131	-	-
0.9658	7000	0.0195	0.0088	0.4498
0.9796	7100	0.0105	-	-
0.9934	7200	0.0129	-	-
1.0072	7300	0.0355	-	-
1.0210	7400	0.0078	-	-
1.0348	7500	0.0008	-	-
1.0486	7600	0.0004	-	-
1.0624	7700	0.0312	-	-
1.0762	7800	0.0158	-	-
1.0900	7900	0.0153	-	-
1.1038	8000	0.0069	0.0135	0.4659
1.1175	8100	0.0042	-	-
1.1313	8200	0.0071	-	-
1.1451	8300	0.0007	-	-
1.1589	8400	0.0095	-	-
1.1727	8500	0.0212	-	-
1.1865	8600	0.0026	-	-
1.2003	8700	0.0208	-	-
1.2141	8800	0.007	-	-
1.2279	8900	0.0374	-	-
1.2417	9000	0.0026	0.0142	0.4819
1.2555	9100	0.0071	-	-
1.2693	9200	0.0111	-	-
1.2831	9300	0.001	-	-
1.2969	9400	0.0066	-	-
1.3107	9500	0.0065	-	-
1.3245	9600	0.0001	-	-
1.3383	9700	0.0057	-	-
1.3521	9800	0.0162	-	-
1.3659	9900	0.0306	-	-
1.3797	10000	0.0001	0.0058	0.4763
1.3935	10100	0.0002	-	-
1.4073	10200	0.0041	-	-
1.4211	10300	0.0093	-	-
1.4349	10400	0.0075	-	-
1.4487	10500	0.0014	-	-
1.4625	10600	0.0108	-	-
1.4763	10700	0.0014	-	-
1.4901	10800	0.0012	-	-
1.5039	10900	0.0214	-	-
1.5177	11000	0.0018	0.0045	0.4908
1.5315	11100	0.0265	-	-
1.5453	11200	0.0735	-	-
1.5591	11300	0.0039	-	-
1.5728	11400	0.0079	-	-
1.5866	11500	0.0	-	-
1.6004	11600	0.0229	-	-
1.6142	11700	0.0025	-	-
1.6280	11800	0.0152	-	-
1.6418	11900	0.0092	-	-
1.6556	12000	0.0	0.0110	0.4794
1.6694	12100	0.0007	-	-
1.6832	12200	0.0237	-	-
1.6970	12300	0.0062	-	-
1.7108	12400	0.0006	-	-
1.7246	12500	0.0021	-	-
1.7384	12600	0.0241	-	-
1.7522	12700	0.0062	-	-
1.7660	12800	0.0021	-	-
1.7798	12900	0.0012	-	-
1.7936	13000	0.0021	0.0094	0.4797
1.8074	13100	0.0018	-	-
1.8212	13200	0.0009	-	-
1.8350	13300	0.0066	-	-
1.8488	13400	0.0007	-	-
1.8626	13500	0.0116	-	-
1.8764	13600	0.0002	-	-
1.8902	13700	0.0004	-	-
1.9040	13800	0.0116	-	-
1.9178	13900	0.0148	-	-
1.9316	14000	0.0052	0.0118	0.4802
1.9454	14100	0.0	-	-
1.9592	14200	0.0001	-	-
1.9730	14300	0.0215	-	-
1.9868	14400	0.0001	-	-
2.0006	14500	0.0075	-	-
2.0143	14600	0.0063	-	-
2.0281	14700	0.0	-	-
2.0419	14800	0.0102	-	-
2.0557	14900	0.0065	-	-
2.0695	15000	0.0022	0.0102	0.4827
2.0833	15100	0.0019	-	-
2.0971	15200	0.0173	-	-
2.1109	15300	0.0144	-	-
2.1247	15400	0.0079	-	-
2.1385	15500	0.0114	-	-
2.1523	15600	0.0197	-	-
2.1661	15700	0.0263	-	-
2.1799	15800	0.0155	-	-
2.1937	15900	0.011	-	-
2.2075	16000	0.0166	0.0050	0.4842
2.2213	16100	0.0145	-	-
2.2351	16200	0.0001	-	-
2.2489	16300	0.0211	-	-
2.2627	16400	0.0061	-	-
2.2765	16500	0.0109	-	-
2.2903	16600	0.0006	-	-
2.3041	16700	0.0315	-	-
2.3179	16800	0.0089	-	-
2.3317	16900	0.0098	-	-
2.3455	17000	0.008	0.0055	0.4894
2.3593	17100	0.0284	-	-
2.3731	17200	0.0378	-	-
2.3869	17300	0.0058	-	-
2.4007	17400	0.0015	-	-
2.4145	17500	0.0074	-	-
2.4283	17600	0.014	-	-
2.4421	17700	0.0016	-	-
2.4558	17800	0.0049	-	-
2.4696	17900	0.0119	-	-
2.4834	18000	0.0005	0.0054	0.4597
2.4972	18100	0.0069	-	-
2.5110	18200	0.0005	-	-
2.5248	18300	0.0072	-	-
2.5386	18400	0.0321	-	-
2.5524	18500	0.033	-	-
2.5662	18600	0.007	-	-
2.5800	18700	0.0001	-	-
2.5938	18800	0.0021	-	-
2.6076	18900	0.0126	-	-
2.6214	19000	0.0163	0.0038	0.4908
2.6352	19100	0.0149	-	-
2.6490	19200	0.0081	-	-
2.6628	19300	0.0026	-	-
2.6766	19400	0.0002	-	-
2.6904	19500	0.0075	-	-
2.7042	19600	0.0	-	-
2.7180	19700	0.007	-	-
2.7318	19800	0.007	-	-
2.7456	19900	0.0001	-	-
2.7594	20000	0.0057	0.0037	0.4844
2.7732	20100	0.0002	-	-
2.7870	20200	0.01	-	-
2.8008	20300	0.0286	-	-
2.8146	20400	0.0123	-	-
2.8284	20500	0.005	-	-
2.8422	20600	0.0057	-	-
2.8560	20700	0.0028	-	-
2.8698	20800	0.003	-	-
2.8836	20900	0.0046	-	-
2.8974	21000	0.0302	0.0055	0.4845
2.9111	21100	0.0055	-	-
2.9249	21200	0.018	-	-
2.9387	21300	0.0129	-	-
2.9525	21400	0.0079	-	-
2.9663	21500	0.0	-	-
2.9801	21600	0.0003	-	-
2.9939	21700	0.0122	-	-
3.0077	21800	0.0024	-	-
3.0215	21900	0.0028	-	-
3.0353	22000	0.0002	0.0039	0.5145
3.0491	22100	0.0049	-	-
3.0629	22200	0.0027	-	-
3.0767	22300	0.0055	-	-
3.0905	22400	0.0	-	-
3.1043	22500	0.0089	-	-
3.1181	22600	0.0073	-	-
3.1319	22700	0.008	-	-
3.1457	22800	0.0048	-	-
3.1595	22900	0.009	-	-
3.1733	23000	0.0001	0.0036	0.5173
3.1871	23100	0.0004	-	-
3.2009	23200	0.0012	-	-
3.2147	23300	0.0069	-	-
3.2285	23400	0.0001	-	-
3.2423	23500	0.0046	-	-
3.2561	23600	0.0074	-	-
3.2699	23700	0.0161	-	-
3.2837	23800	0.0183	-	-
3.2975	23900	0.0089	-	-
3.3113	24000	0.0116	0.0026	0.5040
3.3251	24100	0.0019	-	-
3.3389	24200	0.0	-	-
3.3526	24300	0.0195	-	-
3.3664	24400	0.0039	-	-
3.3802	24500	0.0065	-	-
3.3940	24600	0.0253	-	-
3.4078	24700	0.0	-	-
3.4216	24800	0.0086	-	-
3.4354	24900	0.0108	-	-
3.4492	25000	0.0053	0.0047	0.5022
3.4630	25100	0.0143	-	-
3.4768	25200	0.0004	-	-
3.4906	25300	0.0079	-	-
3.5044	25400	0.0028	-	-
3.5182	25500	0.0002	-	-
3.5320	25600	0.0	-	-
3.5458	25700	0.0084	-	-
3.5596	25800	0.0101	-	-
3.5734	25900	0.0028	-	-
3.5872	26000	0.0076	0.0054	0.5104
3.6010	26100	0.0066	-	-
3.6148	26200	0.0067	-	-
3.6286	26300	0.0071	-	-
3.6424	26400	0.0001	-	-
3.6562	26500	0.0141	-	-
3.6700	26600	0.0003	-	-
3.6838	26700	0.0005	-	-
3.6976	26800	0.0084	-	-
3.7114	26900	0.0085	-	-
3.7252	27000	0.0023	0.0043	0.5142
3.7390	27100	0.0095	-	-
3.7528	27200	0.0071	-	-
3.7666	27300	0.0002	-	-
3.7804	27400	0.0068	-	-
3.7942	27500	0.0223	-	-
3.8079	27600	0.0155	-	-
3.8217	27700	0.0073	-	-
3.8355	27800	0.0	-	-
3.8493	27900	0.0076	-	-
3.8631	28000	0.0003	0.0026	0.5144
3.8769	28100	0.0137	-	-
3.8907	28200	0.0087	-	-
3.9045	28300	0.0	-	-
3.9183	28400	0.0207	-	-
3.9321	28500	0.0061	-	-
3.9459	28600	0.0137	-	-
3.9597	28700	0.01	-	-
3.9735	28800	0.0067	-	-
3.9873	28900	0.0004	-	-
4.0011	29000	0.0102	0.0035	0.5214
4.0149	29100	0.0101	-	-
4.0287	29200	0.0001	-	-
4.0425	29300	0.0083	-	-
4.0563	29400	0.0087	-	-
4.0701	29500	0.0159	-	-
4.0839	29600	0.0	-	-
4.0977	29700	0.0002	-	-
4.1115	29800	0.0193	-	-
4.1253	29900	0.0	-	-
4.1391	30000	0.0118	0.0030	0.5250
4.1529	30100	0.0439	-	-
4.1667	30200	0.0013	-	-
4.1805	30300	0.001	-	-
4.1943	30400	0.0037	-	-
4.2081	30500	0.0068	-	-
4.2219	30600	0.0276	-	-
4.2357	30700	0.0074	-	-
4.2494	30800	0.0025	-	-
4.2632	30900	0.0006	-	-
4.2770	31000	0.0	0.0031	0.5205
4.2908	31100	0.0066	-	-
4.3046	31200	0.0015	-	-
4.3184	31300	0.0055	-	-
4.3322	31400	0.0067	-	-
4.3460	31500	0.0124	-	-
4.3598	31600	0.0109	-	-
4.3736	31700	0.0077	-	-
4.3874	31800	0.0372	-	-
4.4012	31900	0.0205	-	-
4.4150	32000	0.0001	0.0032	0.5326
4.4288	32100	0.0068	-	-
4.4426	32200	0.0056	-	-
4.4564	32300	0.0001	-	-
4.4702	32400	0.0089	-	-
4.4840	32500	0.0067	-	-
4.4978	32600	0.0053	-	-
4.5116	32700	0.0004	-	-
4.5254	32800	0.012	-	-
4.5392	32900	0.0002	-	-
4.5530	33000	0.0184	0.0024	0.5281
4.5668	33100	0.0147	-	-
4.5806	33200	0.0009	-	-
4.5944	33300	0.0008	-	-
4.6082	33400	0.0036	-	-
4.6220	33500	0.0059	-	-
4.6358	33600	0.0016	-	-
4.6496	33700	0.0091	-	-
4.6634	33800	0.0172	-	-
4.6772	33900	0.008	-	-
4.6909	34000	0.0	0.0026	0.5268
4.7047	34100	0.0001	-	-
4.7185	34200	0.0	-	-
4.7323	34300	0.0003	-	-
4.7461	34400	0.0074	-	-
4.7599	34500	0.0081	-	-
4.7737	34600	0.0053	-	-
4.7875	34700	0.0001	-	-
4.8013	34800	0.0021	-	-
4.8151	34900	0.0001	-	-
4.8289	35000	0.0	0.0028	0.5366
4.8427	35100	0.0	-	-
4.8565	35200	0.0	-	-
4.8703	35300	0.0001	-	-
4.8841	35400	0.0	-	-
4.8979	35500	0.0	-	-
4.9117	35600	0.0146	-	-
4.9255	35700	0.0	-	-
4.9393	35800	0.0038	-	-
4.9531	35900	0.0061	-	-
4.9669	36000	0.0109	0.0028	0.5344
4.9807	36100	0.0058	-	-
4.9945	36200	0.0015	-	-
5.0083	36300	0.0003	-	-
5.0221	36400	0.0	-	-
5.0359	36500	0.0	-	-
5.0497	36600	0.0067	-	-
5.0635	36700	0.0056	-	-
5.0773	36800	0.0066	-	-
5.0911	36900	0.0055	-	-
5.1049	37000	0.0	0.0026	0.5382
5.1187	37100	0.0	-	-
5.1325	37200	0.0054	-	-
5.1462	37300	0.0139	-	-
5.1600	37400	0.0001	-	-
5.1738	37500	0.0	-	-
5.1876	37600	0.0015	-	-
5.2014	37700	0.0	-	-
5.2152	37800	0.0056	-	-
5.2290	37900	0.0	-	-
5.2428	38000	0.0101	0.0031	0.5434
5.2566	38100	0.0002	-	-
5.2704	38200	0.0004	-	-
5.2842	38300	0.0	-	-
5.2980	38400	0.0	-	-
5.3118	38500	0.0	-	-
5.3256	38600	0.0001	-	-
5.3394	38700	0.0002	-	-
5.3532	38800	0.0072	-	-
5.3670	38900	0.0004	-	-
5.3808	39000	0.0011	0.0027	0.5417
5.3946	39100	0.012	-	-
5.4084	39200	0.009	-	-
5.4222	39300	0.0	-	-
5.4360	39400	0.0102	-	-
5.4498	39500	0.0	-	-
5.4636	39600	0.0029	-	-
5.4774	39700	0.0001	-	-
5.4912	39800	0.0	-	-
5.5050	39900	0.0084	-	-
5.5188	40000	0.0001	0.0024	0.5428
5.5326	40100	0.0	-	-
5.5464	40200	0.0001	-	-
5.5602	40300	0.0003	-	-
5.5740	40400	0.0045	-	-
5.5877	40500	0.0001	-	-
5.6015	40600	0.0	-	-
5.6153	40700	0.0003	-	-
5.6291	40800	0.0	-	-
5.6429	40900	0.0053	-	-
5.6567	41000	0.0001	0.0024	0.5440
5.6705	41100	0.0113	-	-
5.6843	41200	0.0069	-	-
5.6981	41300	0.0169	-	-
5.7119	41400	0.0	-	-
5.7257	41500	0.0035	-	-
5.7395	41600	0.0001	-	-
5.7533	41700	0.0001	-	-
5.7671	41800	0.0066	-	-
5.7809	41900	0.0	-	-
5.7947	42000	0.001	0.0026	0.5372
5.8085	42100	0.0079	-	-
5.8223	42200	0.0001	-	-
5.8361	42300	0.0	-	-
5.8499	42400	0.022	-	-
5.8637	42500	0.0208	-	-
5.8775	42600	0.0001	-	-
5.8913	42700	0.0	-	-
5.9051	42800	0.0	-	-
5.9189	42900	0.0125	-	-
5.9327	43000	0.0004	0.0025	0.5403
5.9465	43100	0.0	-	-
5.9603	43200	0.0036	-	-
5.9741	43300	0.0	-	-
5.9879	43400	0.0067	-	-

Framework Versions

Python: 3.11.13
Sentence Transformers: 4.1.0
Transformers: 4.52.4
PyTorch: 2.6.0+cu124
Accelerate: 1.8.1
Datasets: 3.6.0
Tokenizers: 0.21.2

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}
}