Speedsy/turkish-multilingual-e5-small-32768-colbert-cleaned-data-32bsize-1500

PyLate model based on Speedsy/turkish-multilingual-e5-small-32768

This is a PyLate model finetuned from Speedsy/turkish-multilingual-e5-small-32768 on the train 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: Speedsy/turkish-multilingual-e5-small-32768
• Document Length: 180 tokens
• Query Length: 32 tokens
• Output Dimensionality: 128 tokens
• Similarity Function: MaxSim
• Training Dataset:
  • train
• Language: en

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}) with Transformer model: BertModel 
  (1): Dense({'in_features': 384, 'out_features': 128, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

Usage

First install the PyLate library:

pip install -U pylate

Retrieval

PyLate provides a streamlined interface to index and retrieve documents using ColBERT models. The index leverages the Voyager HNSW index to efficiently handle document embeddings and enable fast retrieval.

Indexing documents

First, load the ColBERT model and initialize the Voyager index, then encode and index your documents:

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 Voyager index
index = indexes.Voyager(
    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:

# To load an index, simply instantiate it with the correct folder/name and without overriding it
index = indexes.Voyager(
    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:

# 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:

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,
)

Evaluation

Metrics

Py Late Information Retrieval

• Dataset: ['NanoDBPedia', 'NanoFiQA2018', 'NanoHotpotQA', 'NanoMSMARCO', 'NanoNQ', 'NanoSCIDOCS']
• Evaluated with pylate.evaluation.pylate_information_retrieval_evaluator.PyLateInformationRetrievalEvaluator

Metric	NanoDBPedia	NanoFiQA2018	NanoHotpotQA	NanoMSMARCO	NanoNQ	NanoSCIDOCS
MaxSim_accuracy@1	0.82	0.32	0.8	0.32	0.56	0.32
MaxSim_accuracy@3	0.88	0.48	0.94	0.54	0.7	0.58
MaxSim_accuracy@5	0.92	0.56	0.94	0.6	0.74	0.68
MaxSim_accuracy@10	0.96	0.64	1.0	0.66	0.8	0.78
MaxSim_precision@1	0.82	0.32	0.8	0.32	0.56	0.32
MaxSim_precision@3	0.62	0.22	0.4733	0.18	0.24	0.2667
MaxSim_precision@5	0.568	0.164	0.3	0.12	0.152	0.228
MaxSim_precision@10	0.502	0.1	0.172	0.066	0.084	0.148
MaxSim_recall@1	0.1021	0.1594	0.4	0.32	0.53	0.0667
MaxSim_recall@3	0.1581	0.3228	0.71	0.54	0.67	0.1627
MaxSim_recall@5	0.2215	0.3935	0.75	0.6	0.7	0.2327
MaxSim_recall@10	0.3533	0.4609	0.86	0.66	0.75	0.3017
MaxSim_ndcg@10	0.6323	0.364	0.7771	0.4895	0.6553	0.2903
MaxSim_mrr@10	0.8633	0.4153	0.8642	0.4342	0.6442	0.4729
MaxSim_map@100	0.4745	0.2995	0.6978	0.4475	0.6205	0.2159

Pylate Custom Nano BEIR

• Dataset: NanoBEIR_mean
• Evaluated with pylate_nano_beir_evaluator.PylateCustomNanoBEIREvaluator

Metric	Value
MaxSim_accuracy@1	0.5233
MaxSim_accuracy@3	0.6867
MaxSim_accuracy@5	0.74
MaxSim_accuracy@10	0.8067
MaxSim_precision@1	0.5233
MaxSim_precision@3	0.3333
MaxSim_precision@5	0.2553
MaxSim_precision@10	0.1787
MaxSim_recall@1	0.263
MaxSim_recall@3	0.4273
MaxSim_recall@5	0.4829
MaxSim_recall@10	0.5643
MaxSim_ndcg@10	0.5348
MaxSim_mrr@10	0.6157
MaxSim_map@100	0.4593

Training Details

Training Dataset

train

• Dataset: train at 1072b6b
• Size: 443,147 training samples
• Columns: query_id, document_ids, and scores
• Approximate statistics based on the first 1000 samples:

  	query_id	document_ids	scores
  type	string	list	list
  details	min: 5 tokens mean: 5.83 tokens max: 6 tokens	size: 32 elements	size: 32 elements

• Samples:

  query_id	document_ids	scores
  817836	['2716076', '6741935', '2681109', '5562684', '3507339', ...]	[1.0, 0.7059561610221863, 0.21702419221401215, 0.38270196318626404, 0.20812414586544037, ...]
  1045170	['5088671', '2953295', '8783471', '4268439', '6339935', ...]	[1.0, 0.6493034362792969, 0.0692221149802208, 0.17963139712810516, 0.6697239875793457, ...]
  1069432	['3724008', '314949', '8657336', '7420456', '879004', ...]	[1.0, 0.3706032931804657, 0.3508036434650421, 0.2823200523853302, 0.17563475668430328, ...]

• Loss: pylate.losses.distillation.Distillation

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• learning_rate: 3e-05
• num_train_epochs: 1
• bf16: True

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 8
• 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-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: 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: True
• fp16: False
• 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
• dispatch_batches: None
• split_batches: 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: batch_sampler
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	NanoDBPedia_MaxSim_ndcg@10	NanoFiQA2018_MaxSim_ndcg@10	NanoHotpotQA_MaxSim_ndcg@10	NanoMSMARCO_MaxSim_ndcg@10	NanoNQ_MaxSim_ndcg@10	NanoSCIDOCS_MaxSim_ndcg@10	NanoBEIR_mean_MaxSim_ndcg@10
0.0014	20	0.0332	-	-	-	-	-	-	-
0.0029	40	0.0297	-	-	-	-	-	-	-
0.0043	60	0.0285	-	-	-	-	-	-	-
0.0058	80	0.0289	-	-	-	-	-	-	-
0.0072	100	0.0278	-	-	-	-	-	-	-
0.0087	120	0.0271	-	-	-	-	-	-	-
0.0101	140	0.0267	-	-	-	-	-	-	-
0.0116	160	0.0269	-	-	-	-	-	-	-
0.0130	180	0.0264	-	-	-	-	-	-	-
0.0144	200	0.0262	-	-	-	-	-	-	-
0.0159	220	0.0259	-	-	-	-	-	-	-
0.0173	240	0.026	-	-	-	-	-	-	-
0.0188	260	0.0251	-	-	-	-	-	-	-
0.0202	280	0.025	-	-	-	-	-	-	-
0.0217	300	0.025	-	-	-	-	-	-	-
0.0231	320	0.0261	-	-	-	-	-	-	-
0.0246	340	0.0249	-	-	-	-	-	-	-
0.0260	360	0.0249	-	-	-	-	-	-	-
0.0274	380	0.0243	-	-	-	-	-	-	-
0.0289	400	0.0245	-	-	-	-	-	-	-
0.0303	420	0.0247	-	-	-	-	-	-	-
0.0318	440	0.0245	-	-	-	-	-	-	-
0.0332	460	0.0234	-	-	-	-	-	-	-
0.0347	480	0.0235	-	-	-	-	-	-	-
0.0361	500	0.0245	0.6352	0.3462	0.7674	0.5245	0.6220	0.2769	0.5287
0.0375	520	0.0244	-	-	-	-	-	-	-
0.0390	540	0.0233	-	-	-	-	-	-	-
0.0404	560	0.0239	-	-	-	-	-	-	-
0.0419	580	0.0232	-	-	-	-	-	-	-
0.0433	600	0.0225	-	-	-	-	-	-	-
0.0448	620	0.0234	-	-	-	-	-	-	-
0.0462	640	0.0245	-	-	-	-	-	-	-
0.0477	660	0.0229	-	-	-	-	-	-	-
0.0491	680	0.0232	-	-	-	-	-	-	-
0.0505	700	0.023	-	-	-	-	-	-	-
0.0520	720	0.0238	-	-	-	-	-	-	-
0.0534	740	0.0239	-	-	-	-	-	-	-
0.0549	760	0.0229	-	-	-	-	-	-	-
0.0563	780	0.0237	-	-	-	-	-	-	-
0.0578	800	0.0236	-	-	-	-	-	-	-
0.0592	820	0.0224	-	-	-	-	-	-	-
0.0607	840	0.0226	-	-	-	-	-	-	-
0.0621	860	0.0225	-	-	-	-	-	-	-
0.0635	880	0.023	-	-	-	-	-	-	-
0.0650	900	0.0232	-	-	-	-	-	-	-
0.0664	920	0.0224	-	-	-	-	-	-	-
0.0679	940	0.0227	-	-	-	-	-	-	-
0.0693	960	0.0231	-	-	-	-	-	-	-
0.0708	980	0.0238	-	-	-	-	-	-	-
0.0722	1000	0.0224	0.6463	0.3673	0.7878	0.5283	0.6466	0.2869	0.5439
0.0737	1020	0.0225	-	-	-	-	-	-	-
0.0751	1040	0.0223	-	-	-	-	-	-	-
0.0765	1060	0.023	-	-	-	-	-	-	-
0.0780	1080	0.0218	-	-	-	-	-	-	-
0.0794	1100	0.0228	-	-	-	-	-	-	-
0.0809	1120	0.0219	-	-	-	-	-	-	-
0.0823	1140	0.0225	-	-	-	-	-	-	-
0.0838	1160	0.0231	-	-	-	-	-	-	-
0.0852	1180	0.0233	-	-	-	-	-	-	-
0.0866	1200	0.0224	-	-	-	-	-	-	-
0.0881	1220	0.0223	-	-	-	-	-	-	-
0.0895	1240	0.0216	-	-	-	-	-	-	-
0.0910	1260	0.0227	-	-	-	-	-	-	-
0.0924	1280	0.0218	-	-	-	-	-	-	-
0.0939	1300	0.0222	-	-	-	-	-	-	-
0.0953	1320	0.0218	-	-	-	-	-	-	-
0.0968	1340	0.0216	-	-	-	-	-	-	-
0.0982	1360	0.0227	-	-	-	-	-	-	-
0.0996	1380	0.0211	-	-	-	-	-	-	-
0.1011	1400	0.022	-	-	-	-	-	-	-
0.1025	1420	0.0211	-	-	-	-	-	-	-
0.1040	1440	0.0219	-	-	-	-	-	-	-
0.1054	1460	0.023	-	-	-	-	-	-	-
0.1069	1480	0.0215	-	-	-	-	-	-	-
0.1083	1500	0.022	0.6323	0.3640	0.7771	0.4895	0.6553	0.2903	0.5348

Framework Versions

• Python: 3.11.12
• Sentence Transformers: 4.0.2
• PyLate: 1.2.0
• Transformers: 4.48.2
• PyTorch: 2.6.0+cu124
• Accelerate: 1.6.0
• Datasets: 3.6.0
• Tokenizers: 0.21.1

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"
}

PyLate

@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}
}