souvickdascmsa019/initial-colbert-ir

PyLate model based on colbert-ir/colbertv2.0

This is a PyLate model finetuned from colbert-ir/colbertv2.0 on the reasonir-data 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: colbert-ir/colbertv2.0
• Document Length: 180 tokens
• Query Length: 32 tokens
• Output Dimensionality: 128 tokens
• Similarity Function: MaxSim
• Training Dataset:
  • reasonir-data
• 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': 768, '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

Col BERTTriplet

• Evaluated with pylate.evaluation.colbert_triplet.ColBERTTripletEvaluator

Metric	Value
accuracy	0.9273

Training Details

Training Dataset

reasonir-data

• Dataset: reasonir-data at 0275f82
• Size: 242,513 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: 26.76 tokens max: 32 tokens	min: 4 tokens mean: 19.89 tokens max: 32 tokens	min: 4 tokens mean: 20.32 tokens max: 32 tokens

• Samples:

  query	positive	negative
  Market analysis suggests that the ongoing trend of increased adoption of renewable energy sources will continue to drive the demand for solar panels in the coming years. According to various reports, the global solar panel market is projected to witness significant growth over the next decade, with some analysts predicting a compound annual growth rate (CAGR) of up to 20%. This growth is expected to be fueled by declining costs of production, government incentives, and growing environmental concerns. However, some experts also caution that the market may experience fluctuations due to trade policies, technological advancements, and changes in government regulations. Despite these challenges, the overall outlook for the solar panel market remains optimistic, with many companies investing heavily in research and development to improve efficiency and reduce costs. As the demand for renewable energy continues to rise, it is likely that the solar panel market will play a significant role in...	Contrary to the market analysis suggesting a compound annual growth rate (CAGR) of up to 20% for the global solar panel market, our financial reports indicate a more modest growth rate of 12% over the next decade. While we agree that declining production costs, government incentives, and growing environmental concerns will drive demand for solar panels, we also believe that trade policies, technological advancements, and changes in government regulations will have a more significant impact on the market than previously anticipated. Our projections suggest that the market will experience fluctuations, with some years experiencing higher growth rates than others. However, we do not anticipate the market to experience the same level of growth as predicted by other analysts. Our research indicates that the market will reach a saturation point, beyond which growth will slow down. Additionally, we believe that the impact of advancements in energy storage technologies on the solar panel marke...	The demand for solar panels has been on the rise in recent years, driven by an increase in environmental awareness and the need for sustainable energy sources. One of the key factors contributing to this growth is the decline in production costs. As technology advances, the cost of manufacturing solar panels has decreased, making them more affordable for consumers. Additionally, governments around the world have implemented policies and incentives to encourage the adoption of renewable energy sources, which has further boosted demand for solar panels. However, the solar panel market is not without its challenges. Trade policies and technological advancements can impact the market, and changes in government regulations can create uncertainty. Despite these challenges, the outlook for the solar panel market remains positive, with many companies investing heavily in research and development to improve efficiency and reduce costs. The development of new technologies, such as bifacial panel...
  As the sun set over the vast savannah, a sense of tranquility washed over the pride of lions. Their tawny coats glistened in the fading light, and the sound of crickets provided a soothing background hum. Nearby, a group of humans, armed with cameras and curiosity, observed the wild animals from a safe distance.	The lions lazed in the shade of a nearby tree, their tawny coats a blur as they basked in the warmth. The visitors watched in awe, clicking away at their cameras to capture the majesty of the wild animals. Crickets provided a constant, soothing background noise as the humans took care to keep a safe distance from the pride.	The city's tree planting initiative has been a huge success, providing a serene oasis in the midst of the bustling metropolis. The sounds of the city – car horns, chatter and crickets – blend together to create a symphony of noise. While many humans have been drawn to the tranquility of the park, others have raised concerns about the integration of urban wildlife.
  Recent advancements in the field of artificial intelligence have led to significant breakthroughs in natural language processing. This has far-reaching implications for various industries, including education, where AI-powered chatbots can enhance student learning experiences by providing personalization and real-time feedback. Moreover, the integration of AI in educational settings can help address issues of accessibility and equity.	The rapid expansion of AI research has yielded substantial progress in natural language processing, allowing for the development of more sophisticated AI-powered tools. In the education sector, AI-driven chatbots can facilitate individualized learning and offer instantaneous feedback, thereby enriching the overall learning environment. However, it is crucial to address concerns surrounding the digital divide to ensure that these technological advancements are accessible to all.	One of the primary challenges facing archaeologists today is the authentication of ancient artifacts, which often involves meticulous analysis of relics and literary texts. The discovery of a previously unknown scroll, buried deep within the labyrinthine passages of an Egyptian tomb, shed new light on the role of language in ancient cultures. Interestingly, the sophisticated syntax and nuanced vocabulary of the ancient Egyptian language have some similarities with modern-day linguistic structures.

• Loss: pylate.losses.contrastive.Contrastive

Evaluation Dataset

reasonir-data

• Dataset: reasonir-data at 0275f82
• Size: 2,450 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: 26.92 tokens max: 32 tokens	min: 4 tokens mean: 19.98 tokens max: 32 tokens	min: 4 tokens mean: 20.5 tokens max: 32 tokens

• Samples:

  query	positive	negative
  In the recent speech, the politician claimed that the current tax reform will benefit the middle class and lead to a significant increase in economic growth. The politician stated, 'Our plan is to cut taxes across the board, simplifying the tax code and making it fairer for all Americans. We're taking the money from the unfair and complex system and putting it back in the pockets of the hardworking people.' The politician also emphasized the effects of the cut taxes, 'When we cut taxes, we're putting more money in the hands of our small business owners and workers, who are the backbone of our economy. This means new jobs will be created, wages will rise, and the people who actually create the jobs, the entrepreneurs, will have the funds needed to invest more into their businesses.' Moreover, the politician asserted, 'Our country will witness a major boost in job creation and economic growth which in turn will positively affect local communities all around the country.' Furthermore, the...	According to reputable sources in economics research, the tax reform that the administration is trying to implement closely resembles that of the 2001 and the 2003 cuts under former President George Bush and that of 1981 under President Ronald Reagan who reduced tax rates 23% and 25% respectively. Research carried out by a major university indicated that these reforms only yielded an estimated 10% increase in tax revenue, since the decrease in tax income could result in compensated revenues through economic stimulation. Some studies were actually pointing to the idea that no trickle-down economics apply as more funds were being placed in the already wealthy communities. This change could shift the economical inequalities to an extreme and showed a direct relationship between a tax reduction and a large national deficit increase. Employer demand for the borderline employee may not actually increase from the creation of the new jobs, and economists believed. The variation of wages for jo...	The concept of a universal basic income has been a topic of discussion among economists and policymakers in recent years. While some see it as a viable solution to poverty and economic inequality, others argue that it is not feasible due to the financial constraints it would impose on governments. One of the main concerns is that implementing a universal basic income would require significant funding, which would likely come from increased taxes or redistribution of existing social welfare funds. Critics argue that this could lead to a decrease in economic growth, as people may be less incentivized to work if they are receiving a guaranteed income. On the other hand, proponents argue that a universal basic income would provide a safety net for the most vulnerable members of society and allow people to pursue meaningful work rather than just taking any job for the sake of a paycheck. Some countries have experimented with universal basic income pilots, but the results have been mixed. Fi...
  Recent advances in super-resolution techniques have led to a greater understanding of many sub-cellular structures and have opened up new avenues for exploring cellular behavior at the nanoscale. Fluorescence imaging, in particular, has greatly benefited from these advances and has enabled researchers to visualize the distribution and dynamics of proteins in real time. However, further developments in fluorescence imaging rely on a better comprehension of the interactions between imaging probes and their molecular environment. A crucial factor in these interactions is the size and shape of the probes, which must be optimized to minimize disruption of the native dynamics of the system while also achieving high fluorescence yields. The DNA-based probes have emerged as a promising solution, offering the opportunity to tune the size and shape of the probes to optimize performance.	Microscopy	Biophysics
  I recently purchased this top-of-the-line smartwatch for my birthday, and I must say that it has been a revelation in terms of keeping track of my vital signs and daily activity levels. The watch has an elegant design that doesn't clash with my other wearable accessories, and I love how the touchscreen display lets me access a wealth of health metrics at a glance. Although I've encountered several instances where the heart rate monitoring system was delayed in capturing accurate readings, this minor shortcoming hardly detracts from my overall satisfaction with the product. The value proposition it presents in terms of quality, accuracy, and ascendancy over competing offerings makes it a compelling option in this class of devices. Despite never having owned one before, I found the smartwatch straightforward to use, and the companion app did an excellent job of simplifying the tracking and analysis of my fitness journey. Nothing in particular distinguishes this product's methodology in c...	authentic	fake

• Loss: pylate.losses.contrastive.Contrastive

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 32
• gradient_accumulation_steps: 2
• learning_rate: 2e-05
• weight_decay: 0.01
• warmup_steps: 100
• fp16: True
• remove_unused_columns: False

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 32
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 2
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 2e-05
• weight_decay: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 3
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• warmup_steps: 100
• 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: False
• 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

Click to expand

Epoch	Step	Training Loss	Validation Loss	accuracy
0.0066	50	3.9255	-	-
0.0132	100	1.7945	-	-
0.0198	150	1.5522	-	-
0.0264	200	1.6267	-	-
0.0330	250	1.5434	-	-
0	0	-	-	0.8714
0.0330	250	-	0.8547	-
0.0396	300	1.3113	-	-
0.0462	350	1.3674	-	-
0.0528	400	1.3417	-	-
0.0594	450	1.2831	-	-
0.0660	500	1.2243	-	-
0	0	-	-	0.8820
0.0660	500	-	0.7873	-
0.0726	550	1.2276	-	-
0.0792	600	1.2502	-	-
0.0858	650	1.2247	-	-
0.0924	700	1.178	-	-
0.0990	750	1.2379	-	-
0	0	-	-	0.8931
0.0990	750	-	0.7503	-
0.1056	800	1.3893	-	-
0.1122	850	1.1852	-	-
0.1187	900	1.1082	-	-
0.1253	950	0.9946	-	-
0.1319	1000	1.1834	-	-
0	0	-	-	0.8996
0.1319	1000	-	0.7309	-
0.1385	1050	1.1556	-	-
0.1451	1100	1.0251	-	-
0.1517	1150	1.1943	-	-
0.1583	1200	1.086	-	-
0.1649	1250	1.1236	-	-
0	0	-	-	0.9008
0.1649	1250	-	0.6946	-
0.1715	1300	1.0485	-	-
0.1781	1350	0.9481	-	-
0.1847	1400	1.0898	-	-
0.1913	1450	1.0863	-	-
0.1979	1500	1.0756	-	-
0	0	-	-	0.9037
0.1979	1500	-	0.6747	-
0.2045	1550	0.9973	-	-
0.2111	1600	1.1098	-	-
0.2177	1650	1.1745	-	-
0.2243	1700	0.9654	-	-
0.2309	1750	1.0919	-	-
0	0	-	-	0.9094
0.2309	1750	-	0.6499	-
0.2375	1800	1.0249	-	-
0.2441	1850	0.9863	-	-
0.2507	1900	1.1091	-	-
0.2573	1950	1.0989	-	-
0.2639	2000	1.0662	-	-
0	0	-	-	0.9065
0.2639	2000	-	0.6661	-
0.2705	2050	1.0456	-	-
0.2771	2100	1.1349	-	-
0.2837	2150	1.0111	-	-
0.2903	2200	1.026	-	-
0.2969	2250	0.9415	-	-
0	0	-	-	0.9073
0.2969	2250	-	0.6390	-
0.3035	2300	0.9761	-	-
0.3101	2350	0.9748	-	-
0.3167	2400	1.0238	-	-
0.3233	2450	1.0456	-	-
0.3299	2500	0.9895	-	-
0	0	-	-	0.9110
0.3299	2500	-	0.6435	-
0.3365	2550	0.8796	-	-
0.3431	2600	1.0172	-	-
0.3497	2650	1.014	-	-
0.3562	2700	0.9748	-	-
0.3628	2750	0.9273	-	-
0	0	-	-	0.9082
0.3628	2750	-	0.6303	-
0.3694	2800	1.0122	-	-
0.3760	2850	1.0054	-	-
0.3826	2900	0.8974	-	-
0.3892	2950	0.9396	-	-
0.3958	3000	0.8734	-	-
0	0	-	-	0.9049
0.3958	3000	-	0.6238	-
0.4024	3050	1.0048	-	-
0.4090	3100	0.9701	-	-
0.4156	3150	0.9924	-	-
0.4222	3200	0.9349	-	-
0.4288	3250	0.974	-	-
0	0	-	-	0.9118
0.4288	3250	-	0.6216	-
0.4354	3300	1.0539	-	-
0.4420	3350	0.9389	-	-
0.4486	3400	0.9171	-	-
0.4552	3450	0.9706	-	-
0.4618	3500	1.0124	-	-
0	0	-	-	0.9065
0.4618	3500	-	0.6126	-
0.4684	3550	0.9215	-	-
0.4750	3600	0.8563	-	-
0.4816	3650	0.8249	-	-
0.4882	3700	0.8794	-	-
0.4948	3750	1.0013	-	-
0	0	-	-	0.9078
0.4948	3750	-	0.6022	-
0.5014	3800	0.922	-	-
0.5080	3850	0.9168	-	-
0.5146	3900	0.9315	-	-
0.5212	3950	0.9265	-	-
0.5278	4000	0.9453	-	-
0	0	-	-	0.9078
0.5278	4000	-	0.6083	-
0.5344	4050	0.9585	-	-
0.5410	4100	0.9886	-	-
0.5476	4150	0.9081	-	-
0.5542	4200	0.8181	-	-
0.5608	4250	0.8806	-	-
0	0	-	-	0.9118
0.5608	4250	-	0.5918	-
0.5674	4300	0.858	-	-
0.5740	4350	0.8781	-	-
0.5806	4400	0.9059	-	-
0.5871	4450	0.8475	-	-
0.5937	4500	0.9649	-	-
0	0	-	-	0.9057
0.5937	4500	-	0.5951	-
0.6003	4550	0.969	-	-
0.6069	4600	0.8685	-	-
0.6135	4650	0.9555	-	-
0.6201	4700	0.9166	-	-
0.6267	4750	0.877	-	-
0	0	-	-	0.9073
0.6267	4750	-	0.5858	-
0.6333	4800	0.938	-	-
0.6399	4850	0.9211	-	-
0.6465	4900	0.9699	-	-
0.6531	4950	0.8818	-	-
0.6597	5000	0.9814	-	-
0	0	-	-	0.9176
0.6597	5000	-	0.5705	-
0.6663	5050	0.8588	-	-
0.6729	5100	0.8922	-	-
0.6795	5150	1.0096	-	-
0.6861	5200	0.9217	-	-
0.6927	5250	0.9446	-	-
0	0	-	-	0.9147
0.6927	5250	-	0.5740	-
0.6993	5300	0.9301	-	-
0.7059	5350	0.8436	-	-
0.7125	5400	0.8547	-	-
0.7191	5450	0.9552	-	-
0.7257	5500	0.9227	-	-
0	0	-	-	0.9135
0.7257	5500	-	0.5913	-
0.7323	5550	0.8813	-	-
0.7389	5600	0.8519	-	-
0.7455	5650	0.8223	-	-
0.7521	5700	0.8603	-	-
0.7587	5750	0.8208	-	-
0	0	-	-	0.9151
0.7587	5750	-	0.5698	-
0.7653	5800	0.8542	-	-
0.7719	5850	0.7924	-	-
0.7785	5900	0.9238	-	-
0.7851	5950	0.8303	-	-
0.7917	6000	0.8254	-	-
0	0	-	-	0.9159
0.7917	6000	-	0.5643	-
0.7983	6050	0.8556	-	-
0.8049	6100	0.9286	-	-
0.8115	6150	0.8776	-	-
0.8180	6200	0.8146	-	-
0.8246	6250	0.8469	-	-
0	0	-	-	0.9127
0.8246	6250	-	0.5627	-
0.8312	6300	0.9719	-	-
0.8378	6350	0.9297	-	-
0.8444	6400	0.896	-	-
0.8510	6450	0.8709	-	-
0.8576	6500	0.9436	-	-
0	0	-	-	0.9159
0.8576	6500	-	0.5638	-
0.8642	6550	0.8938	-	-
0.8708	6600	0.8065	-	-
0.8774	6650	0.8281	-	-
0.8840	6700	0.8449	-	-
0.8906	6750	0.813	-	-
0	0	-	-	0.9167
0.8906	6750	-	0.5694	-
0.8972	6800	0.9052	-	-
0.9038	6850	0.9501	-	-
0.9104	6900	0.9612	-	-
0.9170	6950	0.8649	-	-
0.9236	7000	0.7366	-	-
0	0	-	-	0.9188
0.9236	7000	-	0.5691	-
0.9302	7050	0.9621	-	-
0.9368	7100	0.9154	-	-
0.9434	7150	0.8617	-	-
0.9500	7200	0.8629	-	-
0.9566	7250	0.899	-	-
0	0	-	-	0.9159
0.9566	7250	-	0.5559	-
0.9632	7300	0.7766	-	-
0.9698	7350	0.8968	-	-
0.9764	7400	0.8462	-	-
0.9830	7450	0.8478	-	-
0.9896	7500	0.8184	-	-
0	0	-	-	0.9163
0.9896	7500	-	0.5564	-
0.9962	7550	0.8445	-	-
1.0028	7600	0.7305	-	-
1.0094	7650	0.695	-	-
1.0160	7700	0.779	-	-
1.0226	7750	0.5876	-	-
0	0	-	-	0.9184
1.0226	7750	-	0.5776	-
1.0292	7800	0.6372	-	-
1.0358	7850	0.7066	-	-
1.0424	7900	0.6561	-	-
1.0490	7950	0.6854	-	-
1.0555	8000	0.7083	-	-
0	0	-	-	0.9212
1.0555	8000	-	0.5645	-
1.0621	8050	0.6618	-	-
1.0687	8100	0.6602	-	-
1.0753	8150	0.7141	-	-
1.0819	8200	0.7599	-	-
1.0885	8250	0.6307	-	-
0	0	-	-	0.9159
1.0885	8250	-	0.5608	-
1.0951	8300	0.6611	-	-
1.1017	8350	0.6308	-	-
1.1083	8400	0.7035	-	-
1.1149	8450	0.683	-	-
1.1215	8500	0.7077	-	-
0	0	-	-	0.9184
1.1215	8500	-	0.5556	-
1.1281	8550	0.7153	-	-
1.1347	8600	0.6186	-	-
1.1413	8650	0.6289	-	-
1.1479	8700	0.5718	-	-
1.1545	8750	0.5749	-	-
0	0	-	-	0.9167
1.1545	8750	-	0.5695	-
1.1611	8800	0.6788	-	-
1.1677	8850	0.7731	-	-
1.1743	8900	0.6954	-	-
1.1809	8950	0.7404	-	-
1.1875	9000	0.6871	-	-
0	0	-	-	0.9208
1.1875	9000	-	0.5666	-
1.1941	9050	0.6415	-	-
1.2007	9100	0.6517	-	-
1.2073	9150	0.7354	-	-
1.2139	9200	0.7325	-	-
1.2205	9250	0.6272	-	-
0	0	-	-	0.9147
1.2205	9250	-	0.5714	-
1.2271	9300	0.7292	-	-
1.2337	9350	0.6325	-	-
1.2403	9400	0.6344	-	-
1.2469	9450	0.7218	-	-
1.2535	9500	0.6815	-	-
0	0	-	-	0.9176
1.2535	9500	-	0.5651	-
1.2601	9550	0.7186	-	-
1.2667	9600	0.6145	-	-
1.2733	9650	0.7095	-	-
1.2799	9700	0.674	-	-
1.2864	9750	0.7405	-	-
0	0	-	-	0.9200
1.2864	9750	-	0.5666	-
1.2930	9800	0.7186	-	-
1.2996	9850	0.6352	-	-
1.3062	9900	0.7077	-	-
1.3128	9950	0.6873	-	-
1.3194	10000	0.5939	-	-
0	0	-	-	0.9204
1.3194	10000	-	0.5752	-
1.3260	10050	0.7171	-	-
1.3326	10100	0.6592	-	-
1.3392	10150	0.6631	-	-
1.3458	10200	0.7658	-	-
1.3524	10250	0.6213	-	-
0	0	-	-	0.9180
1.3524	10250	-	0.5678	-
1.3590	10300	0.6486	-	-
1.3656	10350	0.662	-	-
1.3722	10400	0.6924	-	-
1.3788	10450	0.7106	-	-
1.3854	10500	0.7239	-	-
0	0	-	-	0.9184
1.3854	10500	-	0.5687	-
1.3920	10550	0.735	-	-
1.3986	10600	0.6784	-	-
1.4052	10650	0.6886	-	-
1.4118	10700	0.649	-	-
1.4184	10750	0.6133	-	-
0	0	-	-	0.9200
1.4184	10750	-	0.5683	-
1.4250	10800	0.6635	-	-
1.4316	10850	0.6803	-	-
1.4382	10900	0.6497	-	-
1.4448	10950	0.6812	-	-
1.4514	11000	0.7493	-	-
0	0	-	-	0.9220
1.4514	11000	-	0.5587	-
1.4580	11050	0.6694	-	-
1.4646	11100	0.6782	-	-
1.4712	11150	0.6839	-	-
1.4778	11200	0.671	-	-
1.4844	11250	0.7648	-	-
0	0	-	-	0.9208
1.4844	11250	-	0.5466	-
1.4910	11300	0.7448	-	-
1.4976	11350	0.6811	-	-
1.5042	11400	0.6984	-	-
1.5108	11450	0.6676	-	-
1.5174	11500	0.7054	-	-
0	0	-	-	0.9204
1.5174	11500	-	0.5569	-
1.5239	11550	0.6109	-	-
1.5305	11600	0.7581	-	-
1.5371	11650	0.7035	-	-
1.5437	11700	0.6943	-	-
1.5503	11750	0.6225	-	-
0	0	-	-	0.9224
1.5503	11750	-	0.5571	-
1.5569	11800	0.661	-	-
1.5635	11850	0.635	-	-
1.5701	11900	0.613	-	-
1.5767	11950	0.6502	-	-
1.5833	12000	0.6935	-	-
0	0	-	-	0.9200
1.5833	12000	-	0.5579	-
1.5899	12050	0.6147	-	-
1.5965	12100	0.6575	-	-
1.6031	12150	0.6837	-	-
1.6097	12200	0.7437	-	-
1.6163	12250	0.6808	-	-
0	0	-	-	0.9204
1.6163	12250	-	0.5507	-
1.6229	12300	0.6698	-	-
1.6295	12350	0.6803	-	-
1.6361	12400	0.676	-	-
1.6427	12450	0.6418	-	-
1.6493	12500	0.6042	-	-
0	0	-	-	0.9188
1.6493	12500	-	0.5563	-
1.6559	12550	0.7139	-	-
1.6625	12600	0.6995	-	-
1.6691	12650	0.6097	-	-
1.6757	12700	0.6407	-	-
1.6823	12750	0.5994	-	-
0	0	-	-	0.9249
1.6823	12750	-	0.5621	-
1.6889	12800	0.6642	-	-
1.6955	12850	0.6198	-	-
1.7021	12900	0.6648	-	-
1.7087	12950	0.5644	-	-
1.7153	13000	0.6531	-	-
0	0	-	-	0.9241
1.7153	13000	-	0.5617	-
1.7219	13050	0.6159	-	-
1.7285	13100	0.7855	-	-
1.7351	13150	0.6307	-	-
1.7417	13200	0.61	-	-
1.7483	13250	0.6672	-	-
0	0	-	-	0.9237
1.7483	13250	-	0.5589	-
1.7548	13300	0.6002	-	-
1.7614	13350	0.6638	-	-
1.7680	13400	0.6112	-	-
1.7746	13450	0.6236	-	-
1.7812	13500	0.6245	-	-
0	0	-	-	0.9220
1.7812	13500	-	0.5580	-
1.7878	13550	0.7146	-	-
1.7944	13600	0.5969	-	-
1.8010	13650	0.7246	-	-
1.8076	13700	0.65	-	-
1.8142	13750	0.7136	-	-
0	0	-	-	0.9204
1.8142	13750	-	0.5533	-
1.8208	13800	0.7062	-	-
1.8274	13850	0.6987	-	-
1.8340	13900	0.6642	-	-
1.8406	13950	0.6761	-	-
1.8472	14000	0.6766	-	-
0	0	-	-	0.9212
1.8472	14000	-	0.5655	-
1.8538	14050	0.5758	-	-
1.8604	14100	0.6594	-	-
1.8670	14150	0.7866	-	-
1.8736	14200	0.5798	-	-
1.8802	14250	0.6472	-	-
0	0	-	-	0.9212
1.8802	14250	-	0.5509	-
1.8868	14300	0.7387	-	-
1.8934	14350	0.6677	-	-
1.9000	14400	0.6697	-	-
1.9066	14450	0.6711	-	-
1.9132	14500	0.6988	-	-
0	0	-	-	0.9229
1.9132	14500	-	0.5528	-
1.9198	14550	0.6301	-	-
1.9264	14600	0.6259	-	-
1.9330	14650	0.6223	-	-
1.9396	14700	0.5702	-	-
1.9462	14750	0.6324	-	-
0	0	-	-	0.9253
1.9462	14750	-	0.5508	-
1.9528	14800	0.6409	-	-
1.9594	14850	0.6609	-	-
1.9660	14900	0.6581	-	-
1.9726	14950	0.6313	-	-
1.9792	15000	0.6191	-	-
0	0	-	-	0.9216
1.9792	15000	-	0.5452	-
1.9858	15050	0.6665	-	-
1.9923	15100	0.5907	-	-
1.9989	15150	0.6586	-	-
2.0055	15200	0.5673	-	-
2.0121	15250	0.5516	-	-
0	0	-	-	0.9233
2.0121	15250	-	0.5589	-
2.0187	15300	0.5012	-	-
2.0253	15350	0.5227	-	-
2.0319	15400	0.4449	-	-
2.0385	15450	0.4862	-	-
2.0451	15500	0.5413	-	-
0	0	-	-	0.9233
2.0451	15500	-	0.5642	-
2.0517	15550	0.5462	-	-
2.0583	15600	0.5318	-	-
2.0649	15650	0.5706	-	-
2.0715	15700	0.5055	-	-
2.0781	15750	0.6141	-	-
0	0	-	-	0.9233
2.0781	15750	-	0.5611	-
2.0847	15800	0.5247	-	-
2.0913	15850	0.4817	-	-
2.0979	15900	0.4599	-	-
2.1045	15950	0.5676	-	-
2.1111	16000	0.3992	-	-
0	0	-	-	0.9237
2.1111	16000	-	0.5720	-
2.1177	16050	0.5337	-	-
2.1243	16100	0.4641	-	-
2.1309	16150	0.5636	-	-
2.1375	16200	0.4811	-	-
2.1441	16250	0.499	-	-
0	0	-	-	0.9216
2.1441	16250	-	0.5673	-
2.1507	16300	0.5822	-	-
2.1573	16350	0.5935	-	-
2.1639	16400	0.5028	-	-
2.1705	16450	0.5118	-	-
2.1771	16500	0.5623	-	-
0	0	-	-	0.9261
2.1771	16500	-	0.5656	-
2.1837	16550	0.481	-	-
2.1903	16600	0.5461	-	-
2.1969	16650	0.5802	-	-
2.2035	16700	0.5269	-	-
2.2101	16750	0.5022	-	-
0	0	-	-	0.9220
2.2101	16750	-	0.5671	-
2.2167	16800	0.5203	-	-
2.2232	16850	0.5461	-	-
2.2298	16900	0.5711	-	-
2.2364	16950	0.5615	-	-
2.2430	17000	0.5748	-	-
0	0	-	-	0.9257
2.2430	17000	-	0.5605	-
2.2496	17050	0.5272	-	-
2.2562	17100	0.4401	-	-
2.2628	17150	0.5158	-	-
2.2694	17200	0.5163	-	-
2.2760	17250	0.5195	-	-
0	0	-	-	0.9237
2.2760	17250	-	0.5647	-
2.2826	17300	0.5235	-	-
2.2892	17350	0.5335	-	-
2.2958	17400	0.4915	-	-
2.3024	17450	0.4915	-	-
2.3090	17500	0.4959	-	-
0	0	-	-	0.9233
2.3090	17500	-	0.5675	-
2.3156	17550	0.5161	-	-
2.3222	17600	0.4944	-	-
2.3288	17650	0.5052	-	-
2.3354	17700	0.4937	-	-
2.3420	17750	0.4695	-	-
0	0	-	-	0.9253
2.3420	17750	-	0.5615	-
2.3486	17800	0.5159	-	-
2.3552	17850	0.4992	-	-
2.3618	17900	0.5288	-	-
2.3684	17950	0.5247	-	-
2.3750	18000	0.5491	-	-
0	0	-	-	0.9257
2.3750	18000	-	0.5594	-
2.3816	18050	0.5332	-	-
2.3882	18100	0.529	-	-
2.3948	18150	0.5534	-	-
2.4014	18200	0.5595	-	-
2.4080	18250	0.573	-	-
0	0	-	-	0.9261
2.4080	18250	-	0.5610	-
2.4146	18300	0.4859	-	-
2.4212	18350	0.5019	-	-
2.4278	18400	0.4771	-	-
2.4344	18450	0.5062	-	-
2.4410	18500	0.5342	-	-
0	0	-	-	0.9229
2.4410	18500	-	0.5617	-
2.4476	18550	0.5275	-	-
2.4541	18600	0.576	-	-
2.4607	18650	0.5172	-	-
2.4673	18700	0.5127	-	-
2.4739	18750	0.4728	-	-
0	0	-	-	0.9249
2.4739	18750	-	0.5651	-
2.4805	18800	0.4256	-	-
2.4871	18850	0.4493	-	-
2.4937	18900	0.4881	-	-
2.5003	18950	0.4843	-	-
2.5069	19000	0.517	-	-
0	0	-	-	0.9249
2.5069	19000	-	0.5626	-
2.5135	19050	0.5927	-	-
2.5201	19100	0.5687	-	-
2.5267	19150	0.5261	-	-
2.5333	19200	0.5698	-	-
2.5399	19250	0.5593	-	-
0	0	-	-	0.9269
2.5399	19250	-	0.5581	-
2.5465	19300	0.571	-	-
2.5531	19350	0.5606	-	-
2.5597	19400	0.4912	-	-
2.5663	19450	0.4805	-	-
2.5729	19500	0.5324	-	-
0	0	-	-	0.9282
2.5729	19500	-	0.5537	-
2.5795	19550	0.5584	-	-
2.5861	19600	0.508	-	-
2.5927	19650	0.5231	-	-
2.5993	19700	0.557	-	-
2.6059	19750	0.5338	-	-
0	0	-	-	0.9257
2.6059	19750	-	0.5518	-
2.6125	19800	0.5037	-	-
2.6191	19850	0.6057	-	-
2.6257	19900	0.5571	-	-
2.6323	19950	0.5177	-	-
2.6389	20000	0.4946	-	-
0	0	-	-	0.9253
2.6389	20000	-	0.5548	-
2.6455	20050	0.5256	-	-
2.6521	20100	0.5107	-	-
2.6587	20150	0.5988	-	-
2.6653	20200	0.4907	-	-
2.6719	20250	0.4697	-	-
0	0	-	-	0.9269
2.6719	20250	-	0.5566	-
2.6785	20300	0.4897	-	-
2.6851	20350	0.5088	-	-
2.6916	20400	0.5442	-	-
2.6982	20450	0.536	-	-
2.7048	20500	0.551	-	-
0	0	-	-	0.9269
2.7048	20500	-	0.5562	-
2.7114	20550	0.5038	-	-
2.7180	20600	0.502	-	-
2.7246	20650	0.5021	-	-
2.7312	20700	0.5441	-	-
2.7378	20750	0.4818	-	-
0	0	-	-	0.9286
2.7378	20750	-	0.5548	-
2.7444	20800	0.5012	-	-
2.7510	20850	0.5294	-	-
2.7576	20900	0.4674	-	-
2.7642	20950	0.5436	-	-
2.7708	21000	0.4609	-	-
0	0	-	-	0.9269
2.7708	21000	-	0.5538	-
2.7774	21050	0.5015	-	-
2.7840	21100	0.5299	-	-
2.7906	21150	0.4363	-	-
2.7972	21200	0.5018	-	-
2.8038	21250	0.5079	-	-
0	0	-	-	0.9265
2.8038	21250	-	0.5549	-
2.8104	21300	0.4467	-	-
2.8170	21350	0.5769	-	-
2.8236	21400	0.5323	-	-
2.8302	21450	0.4714	-	-
2.8368	21500	0.4491	-	-
0	0	-	-	0.9257
2.8368	21500	-	0.5538	-
2.8434	21550	0.4801	-	-
2.8500	21600	0.5132	-	-
2.8566	21650	0.4542	-	-
2.8632	21700	0.5015	-	-
2.8698	21750	0.4818	-	-
0	0	-	-	0.9278
2.8698	21750	-	0.5554	-
2.8764	21800	0.5078	-	-
2.8830	21850	0.508	-	-
2.8896	21900	0.5331	-	-
2.8962	21950	0.5185	-	-
2.9028	22000	0.4469	-	-
0	0	-	-	0.9265
2.9028	22000	-	0.5551	-
2.9094	22050	0.4762	-	-
2.9160	22100	0.5799	-	-
2.9225	22150	0.4978	-	-
2.9291	22200	0.566	-	-
2.9357	22250	0.5837	-	-
0	0	-	-	0.9269
2.9357	22250	-	0.5532	-
2.9423	22300	0.5401	-	-
2.9489	22350	0.523	-	-
2.9555	22400	0.5913	-	-
2.9621	22450	0.4701	-	-
2.9687	22500	0.5568	-	-
0	0	-	-	0.9273
2.9687	22500	-	0.5529	-
2.9753	22550	0.5266	-	-
2.9819	22600	0.4969	-	-
2.9885	22650	0.4917	-	-
2.9951	22700	0.5128	-	-

Framework Versions

• Python: 3.12.4
• Sentence Transformers: 4.0.2
• PyLate: 1.2.0
• Transformers: 4.48.2
• PyTorch: 2.6.0+cu124
• Accelerate: 1.7.0
• 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}
}