ayushexel/reranker-ms-marco-MiniLM-L6-v2-gooaq-bce-500k

ModernBERT-base trained on GooAQ

This is a Cross Encoder model finetuned from cross-encoder/ms-marco-MiniLM-L6-v2 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: cross-encoder/ms-marco-MiniLM-L6-v2
• Maximum Sequence Length: 512 tokens
• Number of Output Labels: 1 label
• Language: en
• License: apache-2.0

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("ayushexel/reranker-ms-marco-MiniLM-L6-v2-gooaq-bce-500k")
# Get scores for pairs of texts
pairs = [
    ['what are the 50 state mottos?', '[\'Maine. "Dirigo" (\\u200bI direct)\', \'44. California. "\\u200bEureka" (I have found it) ... \', \'Arizona. "Ditat Deus" (\\u200bGod Enriches) ... \', \'Indiana. "The Crossroads of America" ... \', \'Alaska. "North to the Future" ... \', \'Utah. "Industry" ... \', \'Delaware. "Liberty and Independence" ... \', \'Maryland. "Fatti maschii, parole femine" (Manly deeds womanly words) ... \']'],
    ['what does it mean when you have white pee?', 'A milky quality to your urine is typically caused by your body sending an increase in white blood cells to fight an infection. When these white blood exit your body via your urine, the cells mix, and your urine appears cloudy.'],
    ['what does it mean when you have white pee?', 'White balance (WB) is the process of removing unrealistic color casts, so that objects which appear white in person are rendered white in your photo. Proper camera white balance has to take into account the "color temperature" of a light source, which refers to the relative warmth or coolness of white light.'],
    ['what does it mean when you have white pee?', "['Lower abdominal pain.', 'Pain during urination.', 'Frequent urination.', 'Difficulty urinating or interrupted urine flow.', 'Blood in the urine.', 'Cloudy or abnormally dark-colored urine.']"],
    ['what does it mean when you have white pee?', 'Peeps. ... Peeps are marshmallows sold in the United States and Canada that are shaped into chicks, bunnies, and other animals.'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)

# Or rank different texts based on similarity to a single text
ranks = model.rank(
    'what are the 50 state mottos?',
    [
        '[\'Maine. "Dirigo" (\\u200bI direct)\', \'44. California. "\\u200bEureka" (I have found it) ... \', \'Arizona. "Ditat Deus" (\\u200bGod Enriches) ... \', \'Indiana. "The Crossroads of America" ... \', \'Alaska. "North to the Future" ... \', \'Utah. "Industry" ... \', \'Delaware. "Liberty and Independence" ... \', \'Maryland. "Fatti maschii, parole femine" (Manly deeds womanly words) ... \']',
        'A milky quality to your urine is typically caused by your body sending an increase in white blood cells to fight an infection. When these white blood exit your body via your urine, the cells mix, and your urine appears cloudy.',
        'White balance (WB) is the process of removing unrealistic color casts, so that objects which appear white in person are rendered white in your photo. Proper camera white balance has to take into account the "color temperature" of a light source, which refers to the relative warmth or coolness of white light.',
        "['Lower abdominal pain.', 'Pain during urination.', 'Frequent urination.', 'Difficulty urinating or interrupted urine flow.', 'Blood in the urine.', 'Cloudy or abnormally dark-colored urine.']",
        'Peeps. ... Peeps are marshmallows sold in the United States and Canada that are shaped into chicks, bunnies, and other animals.',
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

Evaluation

Metrics

Cross Encoder Reranking

• Dataset: gooaq-dev
• Evaluated with CrossEncoderRerankingEvaluator with these parameters:

  {
      "at_k": 10,
      "always_rerank_positives": false
  }
  

Metric	Value
map	0.5832 (+0.2028)
mrr@10	0.5818 (+0.2114)
ndcg@10	0.6298 (+0.1971)

Cross Encoder Reranking

• Datasets: NanoMSMARCO_R100, NanoNFCorpus_R100 and NanoNQ_R100
• Evaluated with CrossEncoderRerankingEvaluator with these parameters:

  {
      "at_k": 10,
      "always_rerank_positives": true
  }
  

Metric	NanoMSMARCO_R100	NanoNFCorpus_R100	NanoNQ_R100
map	0.4501 (-0.0395)	0.3711 (+0.1101)	0.3764 (-0.0432)
mrr@10	0.4371 (-0.0404)	0.5112 (+0.0114)	0.3779 (-0.0488)
ndcg@10	0.5122 (-0.0282)	0.3773 (+0.0523)	0.4386 (-0.0621)

Cross Encoder Nano BEIR

• Dataset: NanoBEIR_R100_mean
• Evaluated with CrossEncoderNanoBEIREvaluator with these parameters:

  {
      "dataset_names": [
          "msmarco",
          "nfcorpus",
          "nq"
      ],
      "rerank_k": 100,
      "at_k": 10,
      "always_rerank_positives": true
  }
  

Metric	Value
map	0.3992 (+0.0091)
mrr@10	0.4421 (-0.0260)
ndcg@10	0.4427 (-0.0127)

Training Details

Training Dataset

Unnamed Dataset

• Size: 3,648,749 training samples
• Columns: question, answer, and label
• Approximate statistics based on the first 1000 samples:

  	question	answer	label
  type	string	string	int
  details	min: 18 characters mean: 43.54 characters max: 83 characters	min: 53 characters mean: 248.29 characters max: 400 characters	0: ~86.10% 1: ~13.90%

• Samples:

  question	answer	label
  what are the 50 state mottos?	['Maine. "Dirigo" (\u200bI direct)', '44. California. "\u200bEureka" (I have found it) ... ', 'Arizona. "Ditat Deus" (\u200bGod Enriches) ... ', 'Indiana. "The Crossroads of America" ... ', 'Alaska. "North to the Future" ... ', 'Utah. "Industry" ... ', 'Delaware. "Liberty and Independence" ... ', 'Maryland. "Fatti maschii, parole femine" (Manly deeds womanly words) ... ']	1
  what does it mean when you have white pee?	A milky quality to your urine is typically caused by your body sending an increase in white blood cells to fight an infection. When these white blood exit your body via your urine, the cells mix, and your urine appears cloudy.	1
  what does it mean when you have white pee?	White balance (WB) is the process of removing unrealistic color casts, so that objects which appear white in person are rendered white in your photo. Proper camera white balance has to take into account the "color temperature" of a light source, which refers to the relative warmth or coolness of white light.	0

• Loss: BinaryCrossEntropyLoss with these parameters:

  {
      "activation_fn": "torch.nn.modules.linear.Identity",
      "pos_weight": 7
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 2048
• per_device_eval_batch_size: 2048
• learning_rate: 2e-05
• warmup_ratio: 0.1
• seed: 12
• bf16: True
• dataloader_num_workers: 12
• 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: 2048
• per_device_eval_batch_size: 2048
• 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: 3
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• warmup_steps: 0
• log_level: passive
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: True
• save_on_each_node: False
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 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: 12
• 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
• 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	gooaq-dev_ndcg@10	NanoMSMARCO_R100_ndcg@10	NanoNFCorpus_R100_ndcg@10	NanoNQ_R100_ndcg@10	NanoBEIR_R100_mean_ndcg@10
-1	-1	-	0.5865 (+0.1538)	0.6686 (+0.1282)	0.3930 (+0.0680)	0.7599 (+0.2592)	0.6072 (+0.1518)
0.0006	1	2.0734	-	-	-	-	-
0.1122	200	1.4177	-	-	-	-	-
0.2245	400	0.7296	-	-	-	-	-
0.3367	600	0.6662	-	-	-	-	-
0.4489	800	0.6445	-	-	-	-	-
0.5612	1000	0.621	0.6176 (+0.1849)	0.5862 (+0.0458)	0.4371 (+0.1121)	0.4987 (-0.0020)	0.5073 (+0.0520)
0.6734	1200	0.6122	-	-	-	-	-
0.7856	1400	0.6031	-	-	-	-	-
0.8979	1600	0.5944	-	-	-	-	-
1.0101	1800	0.5846	-	-	-	-	-
1.1223	2000	0.5647	0.6222 (+0.1895)	0.5471 (+0.0066)	0.4028 (+0.0778)	0.4703 (-0.0304)	0.4734 (+0.0180)
1.2346	2200	0.5636	-	-	-	-	-
1.3468	2400	0.5587	-	-	-	-	-
1.4590	2600	0.5543	-	-	-	-	-
1.5713	2800	0.5559	-	-	-	-	-
1.6835	3000	0.5496	0.6242 (+0.1915)	0.4842 (-0.0563)	0.3852 (+0.0601)	0.4132 (-0.0874)	0.4275 (-0.0279)
1.7957	3200	0.5426	-	-	-	-	-
1.9080	3400	0.5422	-	-	-	-	-
2.0202	3600	0.5426	-	-	-	-	-
2.1324	3800	0.5311	-	-	-	-	-
2.2447	4000	0.5267	0.6291 (+0.1963)	0.5247 (-0.0158)	0.3832 (+0.0581)	0.4469 (-0.0538)	0.4516 (-0.0038)
2.3569	4200	0.526	-	-	-	-	-
2.4691	4400	0.5255	-	-	-	-	-
2.5814	4600	0.5229	-	-	-	-	-
2.6936	4800	0.5206	-	-	-	-	-
2.8058	5000	0.5196	0.6298 (+0.1971)	0.5122 (-0.0282)	0.3773 (+0.0523)	0.4386 (-0.0621)	0.4427 (-0.0127)
2.9181	5200	0.5261	-	-	-	-	-
-1	-1	-	0.6298 (+0.1971)	0.5122 (-0.0282)	0.3773 (+0.0523)	0.4386 (-0.0621)	0.4427 (-0.0127)

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.11.0
• Sentence Transformers: 4.0.1
• Transformers: 4.50.3
• PyTorch: 2.6.0+cu124
• Accelerate: 1.5.2
• Datasets: 3.5.0
• 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",
}