This is a PyLate model finetuned from ozayezerceli/ettin-encoder-32M-TR on the msmarco-tr dataset. It maps sentences & paragraphs to sequences of 128-dimensional dense vectors and can be used for semantic textual similarity using the MaxSim operator.

Model Details

Model Description

Model Type: PyLate model
Base model: ozayezerceli/ettin-encoder-32M-TR
Document Length: 180 tokens
Query Length: 32 tokens
Output Dimensionality: 128 tokens
Similarity Function: MaxSim
Training Dataset:
- msmarco-tr
Language: tr

Model Sources

Documentation: PyLate Documentation
Repository: PyLate on GitHub
Hugging Face: PyLate models on Hugging Face

Full Model Architecture

ColBERT(
  (0): Transformer({'max_seq_length': 179, 'do_lower_case': False, 'architecture': 'ModernBertModel'})
  (1): Dense({'in_features': 384, 'out_features': 128, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity', 'use_residual': False})
)

Usage

First install the PyLate library:

pip install -U pylate

Retrieval

Use this model with PyLate to index and retrieve documents. The index uses FastPLAID for efficient similarity search.

Indexing documents

Load the ColBERT model and initialize the PLAID index, then encode and index your documents:

python

from pylate import indexes, models, retrieve

# Step 1: Load the ColBERT model
model = models.ColBERT(
    model_name_or_path="pylate_model_id",
)

# Step 2: Initialize the PLAID index
index = indexes.PLAID(
    index_folder="pylate-index",
    index_name="index",
    override=True,  # This overwrites the existing index if any
)

# Step 3: Encode the documents
documents_ids = ["1", "2", "3"]
documents = ["document 1 text", "document 2 text", "document 3 text"]

documents_embeddings = model.encode(
    documents,
    batch_size=32,
    is_query=False,  # Ensure that it is set to False to indicate that these are documents, not queries
    show_progress_bar=True,
)

# Step 4: Add document embeddings to the index by providing embeddings and corresponding ids
index.add_documents(
    documents_ids=documents_ids,
    documents_embeddings=documents_embeddings,
)

Note that you do not have to recreate the index and encode the documents every time. Once you have created an index and added the documents, you can re-use the index later by loading it:

python

# To load an index, simply instantiate it with the correct folder/name and without overriding it
index = indexes.PLAID(
    index_folder="pylate-index",
    index_name="index",
)

Retrieving top-k documents for queries

Once the documents are indexed, you can retrieve the top-k most relevant documents for a given set of queries. To do so, initialize the ColBERT retriever with the index you want to search in, encode the queries and then retrieve the top-k documents to get the top matches ids and relevance scores:

python

# Step 1: Initialize the ColBERT retriever
retriever = retrieve.ColBERT(index=index)

# Step 2: Encode the queries
queries_embeddings = model.encode(
    ["query for document 3", "query for document 1"],
    batch_size=32,
    is_query=True,  #  # Ensure that it is set to False to indicate that these are queries
    show_progress_bar=True,
)

# Step 3: Retrieve top-k documents
scores = retriever.retrieve(
    queries_embeddings=queries_embeddings,
    k=10,  # Retrieve the top 10 matches for each query
)

Reranking

If you only want to use the ColBERT model to perform reranking on top of your first-stage retrieval pipeline without building an index, you can simply use rank function and pass the queries and documents to rerank:

python

from pylate import rank, models

queries = [
    "query A",
    "query B",
]

documents = [
    ["document A", "document B"],
    ["document 1", "document C", "document B"],
]

documents_ids = [
    [1, 2],
    [1, 3, 2],
]

model = models.ColBERT(
    model_name_or_path="pylate_model_id",
)

queries_embeddings = model.encode(
    queries,
    is_query=True,
)

documents_embeddings = model.encode(
    documents,
    is_query=False,
)

reranked_documents = rank.rerank(
    documents_ids=documents_ids,
    queries_embeddings=queries_embeddings,
    documents_embeddings=documents_embeddings,
)

Training Details

Training Dataset

msmarco-tr

Dataset: msmarco-tr at ffad30a
Size: 910,904 training samples
Columns: query, positive, and negative
Approximate statistics based on the first 1000 samples:
query positive negative
type string string string
details
min: 6 tokens
mean: 17.07 tokens
max: 32 tokens
min: 25 tokens
mean: 31.99 tokens
max: 32 tokens
min: 21 tokens
mean: 31.99 tokens
max: 32 tokens

Samples:

query	positive	negative
`sinir dokusundaki miyelin kılıfı nerede`	`Miyelin, bir tabaka oluşturan akson dielektrik (elektriksel olarak yalıtkan) malzemeyi çevreleyen yağlı bir beyaz maddedir, miyelin kılıfı, genellikle sadece bir nöronun aksonu etrafında bulunur. Sinir sisteminin düzgün çalışması için gereklidir. Bir tür glial hücrenin bir dış büyümesidir. Miyelin kılıfının üretimi miyelinasyon olarak adlandırılır. İnsanlarda, miyelin kılıfı 14'üncü haftada başlar.`	`İnsanlarda, dört temel doku tipi vardır: epitel dokusu, bağ dokusu, kas dokusu ve sinir dokusu. Her genel doku tipi içinde, belirli doku tipleri vardır. Bunu bir futbol takımı gibi düşünün.Her biri sahada kendi 'iş' olan bireysel oyuncular vardır.n insanlar, dört temel doku tipi vardır: epitel dokusu, bağ dokusu, kas dokusu ve sinir dokusu. Bu genel doku tipinde, her bir genel doku tipinde vardır.`
`Okulların Makine Mühendisliğini Sundukları Şeyler`	`Makine Mühendisliği Teknolojisi Dereceleri için Üst Okullar. Pennsylvania Eyalet Üniversitesi - Harrisburg, Purdue Üniversitesi ve Houston Üniversitesi, makine mühendisliği teknolojisi (MET) alanında lisans derecesi sunan üç okuldur. Bu üniversitelerdeki MET programları hakkında daha fazla bilgi edinmek için okumaya devam edin.`	`Mühendis tanımı, motorların veya makinelerin tasarımında, yapımında ve kullanımında veya çeşitli mühendislik dallarından herhangi birinde eğitimli ve yetenekli bir kişi: bir makine mühendisi; bir inşaat mühendisi. Daha fazla bilgi için bkz.`
`kim navigatör karıştırma valfleri taşır`	`BRADLEY THERMOSTATIC MIXING VANAS. Bradley Corporation, armatür ve sıhhi tesisat ürünlerinin üretiminde lider, dört hat üretir. termostatik karıştırma valfleri (TMVs). Bradley Navigator Yüksek Düşük termostatik karıştırma valfleri vardır. Dıştan gelen talebin çok düşük olduğu uygulamalar için idealdir.`	`Hidrolik Valfler. Eaton valfleri, tüm dünyadaki pazarlarda müşterilerimiz için rekabet avantajı sağlar. Geniş bir seçenek yelpazesinde benzersiz kalite sunan yüksek değerli hidrolik valf ürünlerimiz, gerçek endüstri liderlerinin tüm özelliklerini ve performans seviyelerini içerir. Endüstriyel Valfler.`

Loss: pylate.losses.contrastive.Contrastive

Evaluation Dataset

msmarco-tr

Dataset: msmarco-tr at ffad30a
Size: 9,202 evaluation samples
Columns: query, positive, and negative
Approximate statistics based on the first 1000 samples:
query positive negative
type string string string
details
min: 6 tokens
mean: 16.9 tokens
max: 32 tokens
min: 29 tokens
mean: 32.0 tokens
max: 32 tokens
min: 28 tokens
mean: 31.99 tokens
max: 32 tokens

Samples:

query	positive	negative
`Ermin hangi hayvandır`	`1 Aslında ermine kelimesi beyaz kürklü bir hayvanı ifade ederken, sırt üstü kahverengi kürklü ve karnında baş ve beyaz kürklü bireyler için stoat kullanılır.`	`Dünyada kaç hayvan türü var? İşte kaba bir sayım ve bilim adamlarının sayılara nasıl ulaştıklarına dair kısa bir açıklama. Dünyada kaç hayvan türü var? İşte kaba bir sayım ve bilim adamlarının sayılara nasıl ulaştıklarına dair kısa bir açıklama. Kaç hayvan türü var? [object Object] Strauss, Bob.`
`Abacus nereden çıktı`	`Abacus: Kısa Bir Tarih. Abacus, kökeni Yunanca abax veya abakon (masa veya tablet anlamına gelir) kelimelerinden gelen ve muhtemelen kum anlamına gelen Semitik abq kelimesinden kaynaklanan Latince bir kelimedir. Abacus, büyük sayıları saymak için kullanılan birçok sayma cihazından biridir.`	`Hücre apeksinde, bir flagellum için çapa alanı olan bazal gövdedir. Bazal cisimler, dokuz periferik mikrotübül üçlüsü ile centrioles'inkine benzer bir alt yapıya sahiptir (görüntünün alt merkezindeki yapıya bakınız).`
`Başın arkasında radyasyon tedavisi yüz kızarıklığına neden olur mu`	`Radyasyon Terapisinin En Yaygın Yan Etkileri. Cilt reaksiyonu: Radyasyon tedavisinin yaygın bir yan etkisi, tedavi edilen vücut bölgesinde cilt tahrişidir. Cilt reaksiyonu, hafif kızarıklık ve kuruluktan (güneş yanığına benzer) bazı hastalarda cildin şiddetli soyulmasına (desquamation) kadar değişebilir.`	`Bu açıklama amfizemi işaret edebilir. Bu, sigara içme geçmişiniz varsa daha da muhtemeldir. Radyasyon terapisi bilinen nedenlerden biri değildir. Bu konuda daha fazla cevap almak ve semptomlarınızı çözmeye yardımcı olmak için bir pulmonologla takip etmenizi isteyeceğim. Umarım bu, sorgunuzu tamamen ele alır. Sigara içme geçmişiniz varsa, daha da fazla umut eder. Radyasyon terapisi, bu sorunun çözümüne yardımcı olmanızı ve bu sorunun cevabını takip etmenizi isterim.`

Loss: pylate.losses.contrastive.Contrastive

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 16
per_device_eval_batch_size: 16
learning_rate: 3e-06
num_train_epochs: 1
fp16: True

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
prediction_loss_only: True
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
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: 3e-06
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 1
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
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}
parallelism_config: None
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch_fused
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
hub_revision: None
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
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Click to expand

Epoch	Step	Training Loss
0.0000	1	3.619
0.0018	100	2.4315
0.0035	200	1.6119
0.0053	300	1.304
0.0070	400	1.0853
0.0088	500	0.9767
0.0105	600	0.8725
0.0123	700	0.8096
0.0141	800	0.8027
0.0158	900	0.7689
0.0176	1000	0.73
0.0193	1100	0.7424
0.0211	1200	0.754
0.0228	1300	0.716
0.0246	1400	0.6821
0.0263	1500	0.6501
0.0281	1600	0.6927
0.0299	1700	0.6392
0.0316	1800	0.6635
0.0334	1900	0.6913
0.0351	2000	0.6061
0.0369	2100	0.6164
0.0386	2200	0.6423
0.0404	2300	0.6177
0.0422	2400	0.6503
0.0439	2500	0.5883
0.0457	2600	0.6803
0.0474	2700	0.6367
0.0492	2800	0.5826
0.0509	2900	0.605
0.0527	3000	0.6439
0.0545	3100	0.5931
0.0562	3200	0.5527
0.0580	3300	0.5603
0.0597	3400	0.6135
0.0615	3500	0.5449
0.0632	3600	0.5591
0.0650	3700	0.5143
0.0667	3800	0.5616
0.0685	3900	0.5587
0.0703	4000	0.5455
0.0720	4100	0.59
0.0738	4200	0.5328
0.0755	4300	0.5635
0.0773	4400	0.5446
0.0790	4500	0.5268
0.0808	4600	0.5113
0.0826	4700	0.5601
0.0843	4800	0.5118
0.0861	4900	0.5096
0.0878	5000	0.5721
0.0896	5100	0.5159
0.0913	5200	0.5057
0.0931	5300	0.5588
0.0948	5400	0.5312
0.0966	5500	0.5405
0.0984	5600	0.4901
0.1001	5700	0.4862
0.1019	5800	0.4865
0.1036	5900	0.5623
0.1054	6000	0.5505
0.1071	6100	0.5264
0.1089	6200	0.5081
0.1107	6300	0.5181
0.1124	6400	0.5189
0.1142	6500	0.4838
0.1159	6600	0.4445
0.1177	6700	0.4773
0.1194	6800	0.4977
0.1212	6900	0.5432
0.1230	7000	0.4468
0.1247	7100	0.5046
0.1265	7200	0.5403
0.1282	7300	0.4603
0.1300	7400	0.5605
0.1317	7500	0.5327
0.1335	7600	0.4809
0.1352	7700	0.4764
0.1370	7800	0.475
0.1388	7900	0.4966
0.1405	8000	0.4853
0.1423	8100	0.5495
0.1440	8200	0.4416
0.1458	8300	0.4711
0.1475	8400	0.4707
0.1493	8500	0.4964
0.1511	8600	0.496
0.1528	8700	0.4244
0.1546	8800	0.4495
0.1563	8900	0.5045
0.1581	9000	0.4765
0.1598	9100	0.457
0.1616	9200	0.4489
0.1634	9300	0.4894
0.1651	9400	0.4661
0.1669	9500	0.4305
0.1686	9600	0.4714
0.1704	9700	0.4451
0.1721	9800	0.4976
0.1739	9900	0.3982
0.1756	10000	0.4563
0.1774	10100	0.5073
0.1792	10200	0.4425
0.1809	10300	0.4225
0.1827	10400	0.4651
0.1844	10500	0.4615
0.1862	10600	0.451
0.1879	10700	0.4564
0.1897	10800	0.4465
0.1915	10900	0.4763
0.1932	11000	0.4234
0.1950	11100	0.4708
0.1967	11200	0.4921
0.1985	11300	0.3989
0.2002	11400	0.4292
0.2020	11500	0.4408
0.2038	11600	0.4336
0.2055	11700	0.4584
0.2073	11800	0.4263
0.2090	11900	0.449
0.2108	12000	0.4176
0.2125	12100	0.4277
0.2143	12200	0.449
0.2160	12300	0.4344
0.2178	12400	0.448
0.2196	12500	0.4215
0.2213	12600	0.4207
0.2231	12700	0.4515
0.2248	12800	0.4721
0.2266	12900	0.3955
0.2283	13000	0.3967
0.2301	13100	0.4249
0.2319	13200	0.3995
0.2336	13300	0.4344
0.2354	13400	0.4372
0.2371	13500	0.409
0.2389	13600	0.4563
0.2406	13700	0.4774
0.2424	13800	0.4741
0.2442	13900	0.4025
0.2459	14000	0.4317
0.2477	14100	0.4447
0.2494	14200	0.4075
0.2512	14300	0.3978
0.2529	14400	0.4557
0.2547	14500	0.5149
0.2564	14600	0.5241
0.2582	14700	0.4658
0.2600	14800	0.4291
0.2617	14900	0.4024
0.2635	15000	0.385
0.2652	15100	0.419
0.2670	15200	0.4326
0.2687	15300	0.3958
0.2705	15400	0.4686
0.2723	15500	0.391
0.2740	15600	0.3902
0.2758	15700	0.4507
0.2775	15800	0.4086
0.2793	15900	0.4593
0.2810	16000	0.3708
0.2828	16100	0.3855
0.2845	16200	0.4328
0.2863	16300	0.4165
0.2881	16400	0.4213
0.2898	16500	0.4252
0.2916	16600	0.4378
0.2933	16700	0.3989
0.2951	16800	0.4109
0.2968	16900	0.3761
0.2986	17000	0.4226
0.3004	17100	0.3868
0.3021	17200	0.3842
0.3039	17300	0.3906
0.3056	17400	0.4098
0.3074	17500	0.4119
0.3091	17600	0.4069
0.3109	17700	0.4371
0.3127	17800	0.4051
0.3144	17900	0.4056
0.3162	18000	0.4019
0.3179	18100	0.4082
0.3197	18200	0.4271
0.3214	18300	0.327
0.3232	18400	0.4342
0.3249	18500	0.3808
0.3267	18600	0.3694
0.3285	18700	0.4273
0.3302	18800	0.4286
0.3320	18900	0.3844
0.3337	19000	0.3815
0.3355	19100	0.3941
0.3372	19200	0.3767
0.3390	19300	0.3954
0.3408	19400	0.3917
0.3425	19500	0.3736
0.3443	19600	0.3689
0.3460	19700	0.4003
0.3478	19800	0.3904
0.3495	19900	0.4236
0.3513	20000	0.3917
0.3531	20100	0.3587
0.3548	20200	0.3901
0.3566	20300	0.4165
0.3583	20400	0.3609
0.3601	20500	0.4269
0.3618	20600	0.3774
0.3636	20700	0.4004
0.3653	20800	0.3942
0.3671	20900	0.4091
0.3689	21000	0.3615
0.3706	21100	0.4067
0.3724	21200	0.3496
0.3741	21300	0.3757
0.3759	21400	0.4088
0.3776	21500	0.4301
0.3794	21600	0.3786
0.3812	21700	0.4224
0.3829	21800	0.4049
0.3847	21900	0.3983
0.3864	22000	0.3848
0.3882	22100	0.3807
0.3899	22200	0.3476
0.3917	22300	0.4042
0.3935	22400	0.3554
0.3952	22500	0.409
0.3970	22600	0.3966
0.3987	22700	0.3726
0.4005	22800	0.3709
0.4022	22900	0.3839
0.4040	23000	0.3556
0.4057	23100	0.3789
0.4075	23200	0.3793
0.4093	23300	0.3772
0.4110	23400	0.3775
0.4128	23500	0.3532
0.4145	23600	0.414
0.4163	23700	0.3801
0.4180	23800	0.4054
0.4198	23900	0.3479
0.4216	24000	0.4083
0.4233	24100	0.36
0.4251	24200	0.3935
0.4268	24300	0.3503
0.4286	24400	0.3264
0.4303	24500	0.4038
0.4321	24600	0.3518
0.4339	24700	0.3463
0.4356	24800	0.3509
0.4374	24900	0.3425
0.4391	25000	0.3336
0.4409	25100	0.4048
0.4426	25200	0.3399
0.4444	25300	0.3854
0.4461	25400	0.3817
0.4479	25500	0.3507
0.4497	25600	0.3793
0.4514	25700	0.3534
0.4532	25800	0.3893
0.4549	25900	0.3581
0.4567	26000	0.3576
0.4584	26100	0.3706
0.4602	26200	0.3781
0.4620	26300	0.3886
0.4637	26400	0.3205
0.4655	26500	0.3832
0.4672	26600	0.4126
0.4690	26700	0.3276
0.4707	26800	0.3718
0.4725	26900	0.4142
0.4742	27000	0.3287
0.4760	27100	0.3847
0.4778	27200	0.3567
0.4795	27300	0.372
0.4813	27400	0.384
0.4830	27500	0.3728
0.4848	27600	0.3795
0.4865	27700	0.3653
0.4883	27800	0.3419
0.4901	27900	0.3697
0.4918	28000	0.3441
0.4936	28100	0.3829
0.4953	28200	0.3886
0.4971	28300	0.389
0.4988	28400	0.3833
0.5006	28500	0.3488
0.5024	28600	0.3559
0.5041	28700	0.3922
0.5059	28800	0.3616
0.5076	28900	0.3908
0.5094	29000	0.3875
0.5111	29100	0.3577
0.5129	29200	0.3834
0.5146	29300	0.3792
0.5164	29400	0.3793
0.5182	29500	0.3549
0.5199	29600	0.3363
0.5217	29700	0.3467
0.5234	29800	0.3289
0.5252	29900	0.4189
0.5269	30000	0.3805
0.5287	30100	0.416
0.5305	30200	0.3853
0.5322	30300	0.374
0.5340	30400	0.3798
0.5357	30500	0.3489
0.5375	30600	0.3962
0.5392	30700	0.4032
0.5410	30800	0.3946
0.5428	30900	0.3468
0.5445	31000	0.3582
0.5463	31100	0.3604
0.5480	31200	0.345
0.5498	31300	0.3459
0.5515	31400	0.3461
0.5533	31500	0.3658
0.5550	31600	0.3708
0.5568	31700	0.3546
0.5586	31800	0.3971
0.5603	31900	0.3584
0.5621	32000	0.3197
0.5638	32100	0.3789
0.5656	32200	0.3573
0.5673	32300	0.3439
0.5691	32400	0.3366
0.5709	32500	0.3197
0.5726	32600	0.3508
0.5744	32700	0.4047
0.5761	32800	0.317
0.5779	32900	0.3543
0.5796	33000	0.3923
0.5814	33100	0.346
0.5832	33200	0.3733
0.5849	33300	0.3145
0.5867	33400	0.3408
0.5884	33500	0.4192
0.5902	33600	0.3588
0.5919	33700	0.3377
0.5937	33800	0.3478
0.5954	33900	0.3373
0.5972	34000	0.355
0.5990	34100	0.3262
0.6007	34200	0.3327
0.6025	34300	0.3705
0.6042	34400	0.3229
0.6060	34500	0.3487
0.6077	34600	0.3598
0.6095	34700	0.3499
0.6113	34800	0.3414
0.6130	34900	0.3534
0.6148	35000	0.3292
0.6165	35100	0.3487
0.6183	35200	0.3465
0.6200	35300	0.3653
0.6218	35400	0.3145
0.6236	35500	0.3787
0.6253	35600	0.3302
0.6271	35700	0.3348
0.6288	35800	0.355
0.6306	35900	0.3697
0.6323	36000	0.3532
0.6341	36100	0.3799
0.6358	36200	0.333
0.6376	36300	0.3614
0.6394	36400	0.3268
0.6411	36500	0.3295
0.6429	36600	0.3527
0.6446	36700	0.3267
0.6464	36800	0.3626
0.6481	36900	0.322
0.6499	37000	0.3311
0.6517	37100	0.3336
0.6534	37200	0.3547
0.6552	37300	0.3631
0.6569	37400	0.3328
0.6587	37500	0.3483
0.6604	37600	0.3553
0.6622	37700	0.3419
0.6639	37800	0.368
0.6657	37900	0.3218
0.6675	38000	0.3351
0.6692	38100	0.3906
0.6710	38200	0.3555
0.6727	38300	0.3557
0.6745	38400	0.3411
0.6762	38500	0.329
0.6780	38600	0.3554
0.6798	38700	0.3765
0.6815	38800	0.3867
0.6833	38900	0.3112
0.6850	39000	0.316
0.6868	39100	0.3006
0.6885	39200	0.3202
0.6903	39300	0.3337
0.6921	39400	0.3384
0.6938	39500	0.3845
0.6956	39600	0.3808
0.6973	39700	0.3612
0.6991	39800	0.3269
0.7008	39900	0.3425
0.7026	40000	0.3833
0.7043	40100	0.3548
0.7061	40200	0.3518
0.7079	40300	0.3281
0.7096	40400	0.3627
0.7114	40500	0.3398
0.7131	40600	0.3139
0.7149	40700	0.3155
0.7166	40800	0.341
0.7184	40900	0.3401
0.7202	41000	0.3678
0.7219	41100	0.3134
0.7237	41200	0.32
0.7254	41300	0.3497
0.7272	41400	0.3561
0.7289	41500	0.3501
0.7307	41600	0.3404
0.7325	41700	0.3193
0.7342	41800	0.3517
0.7360	41900	0.3446
0.7377	42000	0.3302
0.7395	42100	0.3384
0.7412	42200	0.3506
0.7430	42300	0.3595
0.7447	42400	0.3088
0.7465	42500	0.338
0.7483	42600	0.3416
0.7500	42700	0.3678
0.7518	42800	0.3949
0.7535	42900	0.3258
0.7553	43000	0.342
0.7570	43100	0.3443
0.7588	43200	0.3364
0.7606	43300	0.3707
0.7623	43400	0.3485
0.7641	43500	0.3374
0.7658	43600	0.2852
0.7676	43700	0.3529
0.7693	43800	0.366
0.7711	43900	0.334
0.7729	44000	0.334
0.7746	44100	0.3827
0.7764	44200	0.3711
0.7781	44300	0.3501
0.7799	44400	0.3291
0.7816	44500	0.3249
0.7834	44600	0.3402
0.7851	44700	0.3452
0.7869	44800	0.3606
0.7887	44900	0.3503
0.7904	45000	0.3513
0.7922	45100	0.3245
0.7939	45200	0.3252
0.7957	45300	0.332
0.7974	45400	0.3306
0.7992	45500	0.3038
0.8010	45600	0.3345
0.8027	45700	0.343
0.8045	45800	0.3069
0.8062	45900	0.3775
0.8080	46000	0.3268
0.8097	46100	0.3168
0.8115	46200	0.2599
0.8133	46300	0.2762
0.8150	46400	0.3322
0.8168	46500	0.3384
0.8185	46600	0.3319
0.8203	46700	0.3151
0.8220	46800	0.3132
0.8238	46900	0.3474
0.8255	47000	0.3414
0.8273	47100	0.3143
0.8291	47200	0.3334
0.8308	47300	0.3312
0.8326	47400	0.322
0.8343	47500	0.3121
0.8361	47600	0.3206
0.8378	47700	0.3384
0.8396	47800	0.3505
0.8414	47900	0.3309
0.8431	48000	0.3456
0.8449	48100	0.3759
0.8466	48200	0.3352
0.8484	48300	0.3063
0.8501	48400	0.3239
0.8519	48500	0.3247
0.8536	48600	0.316
0.8554	48700	0.3099
0.8572	48800	0.3655
0.8589	48900	0.3145
0.8607	49000	0.3206
0.8624	49100	0.3528
0.8642	49200	0.3615
0.8659	49300	0.3213
0.8677	49400	0.3162
0.8695	49500	0.3326
0.8712	49600	0.321
0.8730	49700	0.2965
0.8747	49800	0.3473
0.8765	49900	0.2954
0.8782	50000	0.3059
0.8800	50100	0.3537
0.8818	50200	0.3537
0.8835	50300	0.3764
0.8853	50400	0.2796
0.8870	50500	0.3295
0.8888	50600	0.3075
0.8905	50700	0.3451
0.8923	50800	0.345
0.8940	50900	0.3299
0.8958	51000	0.3451
0.8976	51100	0.3225
0.8993	51200	0.3458
0.9011	51300	0.3225
0.9028	51400	0.3429
0.9046	51500	0.3253
0.9063	51600	0.3442
0.9081	51700	0.314
0.9099	51800	0.314
0.9116	51900	0.362
0.9134	52000	0.3216
0.9151	52100	0.3273
0.9169	52200	0.3118
0.9186	52300	0.3297
0.9204	52400	0.3391
0.9222	52500	0.3739
0.9239	52600	0.3481
0.9257	52700	0.3357
0.9274	52800	0.3202
0.9292	52900	0.3445
0.9309	53000	0.3548
0.9327	53100	0.3385
0.9344	53200	0.3264
0.9362	53300	0.3479
0.9380	53400	0.3402
0.9397	53500	0.3075
0.9415	53600	0.3269
0.9432	53700	0.2983
0.9450	53800	0.3265
0.9467	53900	0.3615
0.9485	54000	0.3459
0.9503	54100	0.3278
0.9520	54200	0.3496
0.9538	54300	0.333
0.9555	54400	0.3234
0.9573	54500	0.3302
0.9590	54600	0.3425
0.9608	54700	0.3263
0.9626	54800	0.3454
0.9643	54900	0.3443
0.9661	55000	0.3343
0.9678	55100	0.3204
0.9696	55200	0.3089
0.9713	55300	0.3572
0.9731	55400	0.3134
0.9748	55500	0.3189
0.9766	55600	0.3195
0.9784	55700	0.3498
0.9801	55800	0.3635
0.9819	55900	0.3368
0.9836	56000	0.3309
0.9854	56100	0.3437
0.9871	56200	0.3375
0.9889	56300	0.331
0.9907	56400	0.3245
0.9924	56500	0.3188
0.9942	56600	0.2976
0.9959	56700	0.3017
0.9977	56800	0.3497
0.9994	56900	0.359

Framework Versions

Python: 3.12.12
Sentence Transformers: 5.1.1
PyLate: 1.3.4
Transformers: 4.56.2
PyTorch: 2.8.0+cu126
Accelerate: 1.10.1
Datasets: 4.0.0
Tokenizers: 0.22.1

Citation

BibTeX

Sentence Transformers

bibtex

@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"
}

PyLate

bibtex

@misc{PyLate,
title={PyLate: Flexible Training and Retrieval for Late Interaction Models},
author={Chaffin, Antoine and Sourty, Raphaël},
url={https://github.com/lightonai/pylate},
year={2024}
}

	query	positive	negative
type	string	string	string
details	min: 6 tokens mean: 17.07 tokens max: 32 tokens	min: 25 tokens mean: 31.99 tokens max: 32 tokens	min: 21 tokens mean: 31.99 tokens max: 32 tokens

newmindai/col-ettin-encoder-32M-TR

PyLate model based on ozayezerceli/ettin-encoder-32M-TR

Model Details

Model Description

Model Sources

Full Model Architecture

Usage

Retrieval

Indexing documents

Retrieving top-k documents for queries

Reranking

Training Details

Training Dataset

msmarco-tr

Evaluation Dataset

msmarco-tr

Training Hyperparameters

Non-Default Hyperparameters

All Hyperparameters

Training Logs

Framework Versions

Citation

BibTeX

Sentence Transformers

PyLate

newmindai/
col-ettin-encoder-32M-TR