bansalaman18/reranker-msmarco-v1.1-ettin-encoder-17m-listnet

CrossEncoder based on jhu-clsp/ettin-encoder-17m

This is a Cross Encoder model finetuned from jhu-clsp/ettin-encoder-17m on the ms_marco dataset using the sentence-transformers library. It computes scores for pairs of texts, which can be used for text reranking and semantic search.

Model Details

Model Description

• Model Type: Cross Encoder
• Base model: jhu-clsp/ettin-encoder-17m
• Maximum Sequence Length: 7999 tokens
• Number of Output Labels: 1 label
• Training Dataset:
  • ms_marco
• Language: en

Model Sources

• Documentation: Sentence Transformers Documentation
• Documentation: Cross Encoder Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Cross Encoders on Hugging Face

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 CrossEncoder

# Download from the 🤗 Hub
model = CrossEncoder("bansalaman18/reranker-msmarco-v1.1-ettin-encoder-17m-listnet")
# Get scores for pairs of texts
pairs = [
    ['what color is dun white sockets mean', 'The color of the monster’s eyes seemed to look match his “dun white sockets” meaning that his eyes have no color, they are just white, and white eyes is something that seems to be so unnatural, so soulless. '],
    ['what color is dun white sockets mean', "The buckskin's color can appear from a very light cream, to yellow, tan, or gold. They will have black points--a black or dark brown color on the tips of the ears, and on their legs, mane and tail. They usually do not have a dorsal tripe. Buckskins are often confused with with dun horses. A buckskin, however, will not have the primitive markings that a dun has (zebra stripes on the legs)."],
    ['what color is dun white sockets mean', 'It cannot have red foals, regardless of the color of the mate. The basic color of the horse will be black, bay or brown, but depending on genes at other color loci, the horse may be buckskin, zebra dun, grullo, perlino, gray, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa. E. No red factor detected. It can transmit either E or e to its offspring. The basic color of the horse will be black, bay or brown, but depending on genes at other color loci, the horse may be buckskin, zebra dun, grullo, perlino, gray, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa.'],
    ['what color is dun white sockets mean', 'The basic color is sorrel or chestnut, but depending on genes at other color loci, the horse could be palomino, red dun, gray, cremello, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa. E/e. Both black and red factors detected. It can transmit either E or e to its offspring. The basic color of the horse will be black, bay or brown, but depending on genes at other color loci, the horse may be buckskin, zebra dun, grullo, perlino, gray, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa. E. No red factor'],
    ['what color is dun white sockets mean', '1 Red dun, also called claybank or fox dun, horses do not have black points, as there is no black on the horse to be affected. 2  Instead, the points and primitive markings are a darker shade of red than the coat. 3  Genetically, the horse has an underlying chestnut coat color, acted upon by the dun gene. 1 Dun, also called bay dun, classic dun or zebra dun, the most common type of dun, has a tan or gold body with black mane, tail, and primitive markings. 2  Genetically, the horse has an underlying bay'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)

# Or rank different texts based on similarity to a single text
ranks = model.rank(
    'what color is dun white sockets mean',
    [
        'The color of the monster’s eyes seemed to look match his “dun white sockets” meaning that his eyes have no color, they are just white, and white eyes is something that seems to be so unnatural, so soulless. ',
        "The buckskin's color can appear from a very light cream, to yellow, tan, or gold. They will have black points--a black or dark brown color on the tips of the ears, and on their legs, mane and tail. They usually do not have a dorsal tripe. Buckskins are often confused with with dun horses. A buckskin, however, will not have the primitive markings that a dun has (zebra stripes on the legs).",
        'It cannot have red foals, regardless of the color of the mate. The basic color of the horse will be black, bay or brown, but depending on genes at other color loci, the horse may be buckskin, zebra dun, grullo, perlino, gray, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa. E. No red factor detected. It can transmit either E or e to its offspring. The basic color of the horse will be black, bay or brown, but depending on genes at other color loci, the horse may be buckskin, zebra dun, grullo, perlino, gray, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa.',
        'The basic color is sorrel or chestnut, but depending on genes at other color loci, the horse could be palomino, red dun, gray, cremello, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa. E/e. Both black and red factors detected. It can transmit either E or e to its offspring. The basic color of the horse will be black, bay or brown, but depending on genes at other color loci, the horse may be buckskin, zebra dun, grullo, perlino, gray, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa. E. No red factor',
        '1 Red dun, also called claybank or fox dun, horses do not have black points, as there is no black on the horse to be affected. 2  Instead, the points and primitive markings are a darker shade of red than the coat. 3  Genetically, the horse has an underlying chestnut coat color, acted upon by the dun gene. 1 Dun, also called bay dun, classic dun or zebra dun, the most common type of dun, has a tan or gold body with black mane, tail, and primitive markings. 2  Genetically, the horse has an underlying bay',
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

Training Details

Training Dataset

ms_marco

• Dataset: ms_marco at a47ee7a
• Size: 78,704 training samples
• Columns: query, docs, and labels
• Approximate statistics based on the first 1000 samples:

  	query	docs	labels
  type	string	list	list
  details	min: 11 characters mean: 33.35 characters max: 99 characters	min: 3 elements mean: 6.50 elements max: 10 elements	min: 3 elements mean: 6.50 elements max: 10 elements

• Samples:

  query	docs	labels
  how soon can the onset of a foodborne illness occur	['1 What some people call the “stomach flu” may actually be a foodborne illness caused by a pathogen (i.e., virus, bacteria, or parasite) in contaminated food or drink. 2 The incubation period (the time between exposure to the pathogen and onset of symptoms) can range from several hours to 1 week. Symptoms of Foodborne Illness. 1 Common symptoms of foodborne illness are diarrhea and/or vomiting, typically lasting 1 to 7 days. 2 Other symptoms might include abdominal cramps, nausea, fever, joint/back aches, and fatigue.', 'The symptoms of botulism make hospitalization necessary. If diagnosed early, foodborne and wound botulism can be treated with an antitoxin that is maintained by the Centers for Disease Control. The antitoxin reduces the progression of paralysis and may reduce the severity and duration of symptoms. How soon after exposure do symptoms occur? Generally, symptoms begin 12-36 hours after eating contaminated food, but may occur as early as a few hours and as late as 10 d...	[1, 0, 0, 0, 0, ...]
  common name of maize plant	['The leafy stalk produces ears which contain the grain, which are seeds called kernels. Maize kernels are often used in cooking as a starch. The six major types of maize are dent, flint, pod, popcorn, flour, and sweet. The word maize derives from the Spanish form of the indigenous Taino taíno word for the, plant. Maiz it is known by other names around the. World the word corn Outside North, America australia And New zealand refers to any cereal, crop its meaning understood to vary geographically to refer to the local. staple', 'corn. This plant can be weedy or invasive according to the authoritative sources noted below.This plant may be known by one or more common names in different places, and some are listed above. Click on an acronym to view each weed list, or click here for a composite list of Weeds of the U.S.', 'Dracaena Corn Plant. The dracaena corn plant (botanical name: dracaena fragrans massangeana) is a well known indoor plant which is grown in many homes and offices within...	[1, 0, 0, 0, 0, ...]
  can i use car oil in my lawn mower	["Yes, you may use car oil in your lawnmower but it is not recommended to use multiweight oil. Use 30 weight only. Reason; the chemicals in multiweight won't cool the aircooled engine as well as 30 weight. I like Castrol 30 weight. There is no difference in oil for a lawn mower and oil for an automobile.", "The wrong kind of oil can ruin your mower's engine. Craftsman offers a line of lawnmowers that includes both push and riding mowers. The engines in these mowers are smaller than car engines, but they work using the same combustion principles. This means they need oil, just as car engines do.", 'Type of Oil. You usually have no need to buy high-priced oil for your mower. SAE 30 motor oil is commonly recommended for use in a lawn mower engine, but the safest best is to use the type of oil your lawn mower manufacturer recommends.', "Usually the type of oil recommended in small gasoline powered engines like a lawn mower is SAE 30. However, you can also use 10W-30 or 10W-40 just fine wit...	[1, 0, 0, 0, 0, ...]

• Loss: ListNetLoss with these parameters:

  {
      "activation_fn": "torch.nn.modules.linear.Identity",
      "mini_batch_size": 16
  }
  

Evaluation Dataset

ms_marco

• Dataset: ms_marco at a47ee7a
• Size: 1,000 evaluation samples
• Columns: query, docs, and labels
• Approximate statistics based on the first 1000 samples:

  	query	docs	labels
  type	string	list	list
  details	min: 11 characters mean: 33.85 characters max: 103 characters	min: 2 elements mean: 6.00 elements max: 10 elements	min: 2 elements mean: 6.00 elements max: 10 elements

• Samples:

  query	docs	labels
  what color is dun white sockets mean	['The color of the monster’s eyes seemed to look match his “dun white sockets” meaning that his eyes have no color, they are just white, and white eyes is something that seems to be so unnatural, so soulless. ', "The buckskin's color can appear from a very light cream, to yellow, tan, or gold. They will have black points--a black or dark brown color on the tips of the ears, and on their legs, mane and tail. They usually do not have a dorsal tripe. Buckskins are often confused with with dun horses. A buckskin, however, will not have the primitive markings that a dun has (zebra stripes on the legs).", 'It cannot have red foals, regardless of the color of the mate. The basic color of the horse will be black, bay or brown, but depending on genes at other color loci, the horse may be buckskin, zebra dun, grullo, perlino, gray, white or any of these colors with the white hair patterns tobiano, overo, roan or appaloosa. E. No red factor detected. It can transmit either E or e to its offspring...	[1, 0, 0, 0, 0, ...]
  what is stainless steel	["Stainless steel is an alloy of Iron with a minimum of 10.5% Chromium. Chromium produces a thin layer of oxide on the surface of the steel known as the 'passive layer' . This prevents any further corrosion of the surface. ", 'In metallurgy, stainless steel, also known as inox steel or inox from French inoxydable , is a steel alloy with a minimum of 10.5% chromium content by mass. Stainless steel does not readily corrode, rust or stain with water as ordinary steel does. However, it is not fully stain-proof in low-oxygen, high-salinity, or poor air-circulation environments. There are different grades and surface finishes of stainless steel to suit the environment the alloy must endure', 'Stainless steel is the universal name for a number of different steels used primarily for their anti-corrosive element. This steel has been developed to resist a number of corrosive environments. It ensures that our workplaces are safe, that buildings last longer and that our food preparation surfaces ...	[1, 1, 0, 0, 0, ...]
  how soon can you introduce baby to almond milk	['Pediatricians will tell you that you should introduce nuts (tree nuts) to your baby between the age of 12 months and 36 months. Oh yes, this is a big and confusing age gap. The reasons for this age gap are as varied as pediatricians that make the recommendations. ', "Report Abuse. no i would not give her almond milk because 1- they aren't supposed to have regular cows milk until 1 year because thats when their digestive systems can handle it and 2- almond is a nut and with how severe nut allergies are, I wouldn't introduce any type of nut until the DR okay's it around 18 months. Report Abuse. I would wait until a year old to introduce any kind of milk. You should tlak to her pediatrician about this. Babies digestive tract is very sensitive at this point, and I'm not sure it would be wise. I don't know much about almond milk though, so don't quote me. I do know that cow's milk can cause intestinal bleeding in infants under 1 year, so almond milk may cause it too", 'Almond milk is easy...	[1, 0, 0, 0, 0, ...]

• Loss: ListNetLoss with these parameters:

  {
      "activation_fn": "torch.nn.modules.linear.Identity",
      "mini_batch_size": 16
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• learning_rate: 2e-05
• num_train_epochs: 5
• seed: 12
• bf16: True
• load_best_model_at_end: 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: 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: 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: 5
• 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: 12
• 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: True
• 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}
• tp_size: 0
• 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: batch_sampler
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss
0.0002	1	2.7003	-
0.0203	100	2.1229	2.0780
0.0407	200	2.0888	2.0754
0.0610	300	2.0847	2.0756
0.0813	400	2.0737	2.0747
0.1016	500	2.0919	2.0728
0.1220	600	2.0737	2.0711
0.1423	700	2.0822	2.0705
0.1626	800	2.0823	2.0703
0.1830	900	2.0767	2.0699
0.2033	1000	2.065	2.0702
0.2236	1100	2.0836	2.0689
0.2440	1200	2.0771	2.0693
0.2643	1300	2.0792	2.0686
0.2846	1400	2.0798	2.0685
0.3049	1500	2.0799	2.0681
0.3253	1600	2.0798	2.0683
0.3456	1700	2.0809	2.0675
0.3659	1800	2.0851	2.0676
0.3863	1900	2.077	2.0675
0.4066	2000	2.0793	2.0672
0.4269	2100	2.0768	2.0671
0.4472	2200	2.0782	2.0667
0.4676	2300	2.0858	2.0669
0.4879	2400	2.0721	2.0667
0.5082	2500	2.0838	2.0662
0.5286	2600	2.0757	2.0664
0.5489	2700	2.0735	2.0658
0.5692	2800	2.0787	2.0663
0.5896	2900	2.0778	2.0660
0.6099	3000	2.0778	2.0657
0.6302	3100	2.0818	2.0654
0.6505	3200	2.077	2.0659
0.6709	3300	2.0808	2.0659
0.6912	3400	2.075	2.0655
0.7115	3500	2.0814	2.0652
0.7319	3600	2.0766	2.0652
0.7522	3700	2.0789	2.0657
0.7725	3800	2.0753	2.0651
0.7928	3900	2.0788	2.0651
0.8132	4000	2.0683	2.0649
0.8335	4100	2.0804	2.0654
0.8538	4200	2.0848	2.0649
0.8742	4300	2.0768	2.0647
0.8945	4400	2.0752	2.0652
0.9148	4500	2.0758	2.0653
0.9351	4600	2.0685	2.0649
0.9555	4700	2.083	2.0647
0.9758	4800	2.0624	2.0648
0.9961	4900	2.079	2.0652
1.0165	5000	2.0784	2.0651
1.0368	5100	2.0663	2.0648
1.0571	5200	2.0721	2.0652
1.0775	5300	2.0813	2.0651
1.0978	5400	2.0784	2.0649
1.1181	5500	2.0782	2.0657
1.1384	5600	2.0626	2.0653
1.1588	5700	2.0679	2.0657
1.1791	5800	2.0718	2.0653
1.1994	5900	2.0772	2.0659
1.2198	6000	2.0676	2.0653
1.2401	6100	2.077	2.0650
1.2604	6200	2.0683	2.0652
1.2807	6300	2.0652	2.0659
1.3011	6400	2.0721	2.0658
1.3214	6500	2.0643	2.0652
1.3417	6600	2.0704	2.0654
1.3621	6700	2.0757	2.0651
1.3824	6800	2.0678	2.0650
1.4027	6900	2.0692	2.0649
1.4231	7000	2.0657	2.0650
1.4434	7100	2.0654	2.0650
1.4637	7200	2.0742	2.0650
1.4840	7300	2.0732	2.0650
1.5044	7400	2.0652	2.0650
1.5247	7500	2.0706	2.0646
1.5450	7600	2.0817	2.0647
1.5654	7700	2.0713	2.0649
1.5857	7800	2.075	2.0651
1.6060	7900	2.0738	2.0655
1.6263	8000	2.076	2.0647
1.6467	8100	2.0658	2.0650
1.6670	8200	2.0759	2.0649
1.6873	8300	2.0713	2.0650
1.7077	8400	2.0746	2.0647
1.7280	8500	2.0727	2.0647
1.7483	8600	2.0714	2.0646
1.7687	8700	2.0752	2.0648
1.7890	8800	2.0744	2.0645
1.8093	8900	2.0677	2.0658
1.8296	9000	2.0786	2.0647
1.8500	9100	2.0648	2.0646
1.8703	9200	2.0578	2.0645
1.8906	9300	2.0725	2.0647
1.911	9400	2.077	2.0644
1.9313	9500	2.0656	2.0644
1.9516	9600	2.0764	2.0647
1.9719	9700	2.0766	2.0647
1.9923	9800	2.0643	2.0646
2.0126	9900	2.0657	2.0651
2.0329	10000	2.0681	2.0658
2.0533	10100	2.0653	2.0655
2.0736	10200	2.0611	2.0684
2.0939	10300	2.0673	2.0673
2.1143	10400	2.0672	2.0680
2.1346	10500	2.074	2.0666
2.1549	10600	2.0536	2.0668
2.1752	10700	2.0708	2.0664
2.1956	10800	2.0604	2.0670
2.2159	10900	2.0684	2.0668
2.2362	11000	2.061	2.0666
2.2566	11100	2.0679	2.0673
2.2769	11200	2.0811	2.0670
2.2972	11300	2.052	2.0668
2.3175	11400	2.0671	2.0666
2.3379	11500	2.0568	2.0666
2.3582	11600	2.066	2.0666
2.3785	11700	2.0696	2.0665
2.3989	11800	2.0602	2.0662
2.4192	11900	2.0599	2.0667
2.4395	12000	2.0703	2.0677
2.4598	12100	2.0684	2.0666
2.4802	12200	2.0673	2.0668
2.5005	12300	2.0722	2.0667
2.5208	12400	2.0666	2.0668
2.5412	12500	2.0742	2.0680
2.5615	12600	2.0619	2.0666
2.5818	12700	2.0618	2.0670
2.6022	12800	2.065	2.0667
2.6225	12900	2.0556	2.0674
2.6428	13000	2.0592	2.0668
2.6631	13100	2.0671	2.0666
2.6835	13200	2.07	2.0664
2.7038	13300	2.063	2.0666
2.7241	13400	2.0665	2.0666
2.7445	13500	2.0664	2.0693
2.7648	13600	2.065	2.0665
2.7851	13700	2.0648	2.0675
2.8054	13800	2.0543	2.0684
2.8258	13900	2.0566	2.0667
2.8461	14000	2.0643	2.0672
2.8664	14100	2.0669	2.0677
2.8868	14200	2.0589	2.0672
2.9071	14300	2.0663	2.0668
2.9274	14400	2.0657	2.0666
2.9478	14500	2.0675	2.0666
2.9681	14600	2.0717	2.0664
2.9884	14700	2.0667	2.0677
3.0087	14800	2.0671	2.0692
3.0291	14900	2.0537	2.0701
3.0494	15000	2.0535	2.0705
3.0697	15100	2.0557	2.0692
3.0901	15200	2.055	2.0700
3.1104	15300	2.0542	2.0714
3.1307	15400	2.0595	2.0705
3.1510	15500	2.0553	2.0709
3.1714	15600	2.051	2.0701
3.1917	15700	2.0583	2.0698
3.2120	15800	2.0679	2.0703
3.2324	15900	2.0566	2.0704
3.2527	16000	2.0554	2.0711
3.2730	16100	2.0623	2.0698
3.2934	16200	2.0597	2.0709
3.3137	16300	2.0558	2.0701
3.3340	16400	2.0657	2.0710
3.3543	16500	2.06	2.0706
3.3747	16600	2.0617	2.0709
3.3950	16700	2.0652	2.0700
3.4153	16800	2.0637	2.0703
3.4357	16900	2.0589	2.0696
3.4560	17000	2.0617	2.0710
3.4763	17100	2.0551	2.0711
3.4966	17200	2.0619	2.0711
3.5170	17300	2.056	2.0719
3.5373	17400	2.0507	2.0705
3.5576	17500	2.0538	2.0706
3.5780	17600	2.0556	2.0708
3.5983	17700	2.0593	2.0716
3.6186	17800	2.0542	2.0714
3.6390	17900	2.0585	2.0709
3.6593	18000	2.0624	2.0706
3.6796	18100	2.0611	2.0705
3.6999	18200	2.0455	2.0708
3.7203	18300	2.0684	2.0701
3.7406	18400	2.0543	2.0701
3.7609	18500	2.0557	2.0713
3.7813	18600	2.0545	2.0720
3.8016	18700	2.0562	2.0703
3.8219	18800	2.059	2.0727
3.8422	18900	2.0592	2.0717
3.8626	19000	2.0533	2.0719
3.8829	19100	2.0595	2.0715
3.9032	19200	2.0575	2.0712
3.9236	19300	2.0608	2.0709
3.9439	19400	2.0561	2.0710
3.9642	19500	2.0619	2.0716
3.9845	19600	2.0585	2.0707
4.0049	19700	2.0606	2.0728
4.0252	19800	2.0556	2.0729
4.0455	19900	2.0461	2.0743
4.0659	20000	2.0509	2.0730
4.0862	20100	2.0597	2.0743
4.1065	20200	2.0493	2.0741
4.1269	20300	2.0522	2.0741
4.1472	20400	2.0542	2.0739
4.1675	20500	2.0591	2.0741
4.1878	20600	2.0453	2.0766
4.2082	20700	2.0493	2.0757
4.2285	20800	2.0472	2.0745
4.2488	20900	2.0524	2.0743
4.2692	21000	2.0522	2.0748
4.2895	21100	2.0515	2.0753
4.3098	21200	2.0487	2.0744
4.3301	21300	2.0555	2.0737
4.3505	21400	2.0456	2.0743
4.3708	21500	2.0552	2.0737
4.3911	21600	2.0531	2.0747
4.4115	21700	2.0562	2.0751
4.4318	21800	2.0496	2.0745
4.4521	21900	2.0512	2.0745
4.4725	22000	2.0536	2.0750
4.4928	22100	2.0577	2.0743
4.5131	22200	2.0499	2.0745
4.5334	22300	2.0567	2.0735
4.5538	22400	2.0574	2.0744
4.5741	22500	2.0438	2.0741
4.5944	22600	2.059	2.0746
4.6148	22700	2.0554	2.0750
4.6351	22800	2.063	2.0743
4.6554	22900	2.0475	2.0741
4.6757	23000	2.0487	2.0739
4.6961	23100	2.0494	2.0752
4.7164	23200	2.0442	2.0751
4.7367	23300	2.0528	2.0743
4.7571	23400	2.0486	2.0746
4.7774	23500	2.0408	2.0747
4.7977	23600	2.0545	2.0749
4.8181	23700	2.0566	2.0752
4.8384	23800	2.0491	2.0746
4.8587	23900	2.0462	2.0744
4.8790	24000	2.0496	2.0749
4.8994	24100	2.0579	2.0743
4.9197	24200	2.0604	2.0743
4.9400	24300	2.0512	2.0744
4.9604	24400	2.0617	2.0745
4.9807	24500	2.0526	2.0746

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.11.13
• Sentence Transformers: 5.0.0
• Transformers: 4.51.0
• PyTorch: 2.9.1+cu126
• Accelerate: 1.8.1
• Datasets: 3.6.0
• Tokenizers: 0.21.4-dev.0

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",
}

ListNetLoss

@inproceedings{cao2007learning,
    title={Learning to Rank: From Pairwise Approach to Listwise Approach},
    author={Cao, Zhe and Qin, Tao and Liu, Tie-Yan and Tsai, Ming-Feng and Li, Hang},
    booktitle={Proceedings of the 24th international conference on Machine learning},
    pages={129--136},
    year={2007}
}