tomaarsen/bert-base-uncased-gooaq-peft

bert-base-uncased adapter finetuned on GooAQ pairs

This is a sentence-transformers model finetuned from google-bert/bert-base-uncased on the gooaq dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: google-bert/bert-base-uncased
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • gooaq
• Language: en
• License: apache-2.0

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)

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 SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("tomaarsen/bert-base-uncased-gooaq-peft")
# Run inference
sentences = [
    'how can i download youtube videos with internet download manager?',
    "['Go to settings and then click on extensions (top left side in chrome).', 'Minimise your browser and open the location (folder) where IDM is installed. ... ', 'Find the file “IDMGCExt. ... ', 'Drag this file to your chrome browser and drop to install the IDM extension.']",
    "Coca-Cola might rot your teeth and load your body with sugar and calories, but it's actually an effective and safe first line of treatment for some stomach blockages, researchers say.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Information Retrieval

• Datasets: NanoClimateFEVER, NanoDBPedia, NanoFEVER, NanoFiQA2018, NanoHotpotQA, NanoMSMARCO, NanoNFCorpus, NanoNQ, NanoQuoraRetrieval, NanoSCIDOCS, NanoArguAna, NanoSciFact and NanoTouche2020
• Evaluated with InformationRetrievalEvaluator

Metric	NanoClimateFEVER	NanoDBPedia	NanoFEVER	NanoFiQA2018	NanoHotpotQA	NanoMSMARCO	NanoNFCorpus	NanoNQ	NanoQuoraRetrieval	NanoSCIDOCS	NanoArguAna	NanoSciFact	NanoTouche2020
cosine_accuracy@1	0.3	0.48	0.6	0.22	0.64	0.22	0.32	0.4	0.84	0.3	0.16	0.38	0.3673
cosine_accuracy@3	0.42	0.78	0.84	0.4	0.74	0.46	0.44	0.54	0.98	0.36	0.52	0.56	0.8571
cosine_accuracy@5	0.48	0.82	0.9	0.5	0.82	0.54	0.46	0.62	0.98	0.54	0.72	0.64	0.9388
cosine_accuracy@10	0.54	0.92	0.96	0.6	0.84	0.68	0.5	0.7	1.0	0.68	0.8	0.7	1.0
cosine_precision@1	0.3	0.48	0.6	0.22	0.64	0.22	0.32	0.4	0.84	0.3	0.16	0.38	0.3673
cosine_precision@3	0.16	0.46	0.28	0.18	0.3133	0.1533	0.2867	0.18	0.3867	0.2	0.1733	0.2	0.4966
cosine_precision@5	0.116	0.416	0.184	0.14	0.224	0.108	0.244	0.124	0.24	0.192	0.144	0.14	0.449
cosine_precision@10	0.066	0.39	0.098	0.098	0.118	0.068	0.178	0.072	0.13	0.142	0.08	0.078	0.3939
cosine_recall@1	0.1483	0.0444	0.59	0.1144	0.32	0.22	0.0229	0.4	0.7573	0.0647	0.16	0.345	0.0307
cosine_recall@3	0.21	0.1092	0.8	0.2189	0.47	0.46	0.0516	0.53	0.9287	0.1247	0.52	0.525	0.1124
cosine_recall@5	0.2567	0.145	0.8567	0.3109	0.56	0.54	0.062	0.59	0.936	0.1967	0.72	0.615	0.1616
cosine_recall@10	0.2867	0.2407	0.9067	0.4079	0.59	0.68	0.0734	0.67	0.9793	0.2907	0.8	0.68	0.2674
cosine_ndcg@10	0.2613	0.4507	0.7556	0.2964	0.5584	0.4416	0.2241	0.5271	0.9154	0.2646	0.4714	0.5211	0.4291
cosine_mrr@10	0.3718	0.6355	0.7192	0.3307	0.7015	0.3667	0.3782	0.4859	0.9053	0.3836	0.3663	0.4848	0.6237
cosine_map@100	0.2163	0.3183	0.7017	0.2334	0.4954	0.3814	0.0878	0.4878	0.889	0.2058	0.3751	0.4707	0.3288

Nano BEIR

• Dataset: NanoBEIR_mean
• Evaluated with NanoBEIREvaluator

Metric	Value
cosine_accuracy@1	0.4021
cosine_accuracy@3	0.6075
cosine_accuracy@5	0.6891
cosine_accuracy@10	0.7631
cosine_precision@1	0.4021
cosine_precision@3	0.2669
cosine_precision@5	0.2093
cosine_precision@10	0.1471
cosine_recall@1	0.2475
cosine_recall@3	0.3893
cosine_recall@5	0.4577
cosine_recall@10	0.5287
cosine_ndcg@10	0.4705
cosine_mrr@10	0.5195
cosine_map@100	0.3993

Training Details

Training Dataset

gooaq

• Dataset: gooaq at b089f72
• Size: 3,012,496 training samples
• Columns: question and answer
• Approximate statistics based on the first 1000 samples:

  	question	answer
  type	string	string
  details	min: 8 tokens mean: 11.86 tokens max: 21 tokens	min: 14 tokens mean: 60.48 tokens max: 138 tokens

• Samples:

  question	answer
  what is the difference between broilers and layers?	An egg laying poultry is called egger or layer whereas broilers are reared for obtaining meat. So a layer should be able to produce more number of large sized eggs, without growing too much. On the other hand, a broiler should yield more meat and hence should be able to grow well.
  what is the difference between chronological order and spatial order?	As a writer, you should always remember that unlike chronological order and the other organizational methods for data, spatial order does not take into account the time. Spatial order is primarily focused on the location. All it does is take into account the location of objects and not the time.
  is kamagra same as viagra?	Kamagra is thought to contain the same active ingredient as Viagra, sildenafil citrate. In theory, it should work in much the same way as Viagra, taking about 45 minutes to take effect, and lasting for around 4-6 hours. However, this will vary from person to person.

• Loss: MatryoshkaLoss with these parameters:

  {
      "loss": "CachedMultipleNegativesRankingLoss",
      "matryoshka_dims": [
          768,
          512,
          256,
          128,
          64,
          32
      ],
      "matryoshka_weights": [
          1,
          1,
          1,
          1,
          1,
          1
      ],
      "n_dims_per_step": -1
  }
  

Evaluation Dataset

gooaq

• Dataset: gooaq at b089f72
• Size: 3,012,496 evaluation samples
• Columns: question and answer
• Approximate statistics based on the first 1000 samples:

  	question	answer
  type	string	string
  details	min: 8 tokens mean: 11.88 tokens max: 22 tokens	min: 14 tokens mean: 61.03 tokens max: 127 tokens

• Samples:

  question	answer
  how do i program my directv remote with my tv?	['Press MENU on your remote.', 'Select Settings & Help > Settings > Remote Control > Program Remote.', 'Choose the device (TV, audio, DVD) you wish to program. ... ', 'Follow the on-screen prompts to complete programming.']
  are rodrigues fruit bats nocturnal?	Before its numbers were threatened by habitat destruction, storms, and hunting, some of those groups could number 500 or more members. Sunrise, sunset. Rodrigues fruit bats are most active at dawn, at dusk, and at night.
  why does your heart rate increase during exercise bbc bitesize?	During exercise there is an increase in physical activity and muscle cells respire more than they do when the body is at rest. The heart rate increases during exercise. The rate and depth of breathing increases - this makes sure that more oxygen is absorbed into the blood, and more carbon dioxide is removed from it.

• Loss: MatryoshkaLoss with these parameters:

  {
      "loss": "CachedMultipleNegativesRankingLoss",
      "matryoshka_dims": [
          768,
          512,
          256,
          128,
          64,
          32
      ],
      "matryoshka_weights": [
          1,
          1,
          1,
          1,
          1,
          1
      ],
      "n_dims_per_step": -1
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 1024
• per_device_eval_batch_size: 1024
• learning_rate: 2e-05
• num_train_epochs: 1
• warmup_ratio: 0.1
• seed: 12
• bf16: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 1024
• per_device_eval_batch_size: 1024
• 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: 1
• 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: 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: False
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	NanoClimateFEVER_cosine_ndcg@10	NanoDBPedia_cosine_ndcg@10	NanoFEVER_cosine_ndcg@10	NanoFiQA2018_cosine_ndcg@10	NanoHotpotQA_cosine_ndcg@10	NanoMSMARCO_cosine_ndcg@10	NanoNFCorpus_cosine_ndcg@10	NanoNQ_cosine_ndcg@10	NanoQuoraRetrieval_cosine_ndcg@10	NanoSCIDOCS_cosine_ndcg@10	NanoArguAna_cosine_ndcg@10	NanoSciFact_cosine_ndcg@10	NanoTouche2020_cosine_ndcg@10	NanoBEIR_mean_cosine_ndcg@10
0	0	-	-	0.1046	0.2182	0.1573	0.0575	0.2597	0.1602	0.0521	0.0493	0.7310	0.1320	0.2309	0.1240	0.0970	0.1826
0.0010	1	28.4479	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0256	25	27.0904	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0512	50	19.016	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0768	75	12.2306	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1024	100	9.0613	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1279	125	7.393	3.7497	0.2787	0.4840	0.7029	0.2589	0.5208	0.4094	0.2117	0.4526	0.9042	0.2503	0.5280	0.4922	0.4132	0.4544
0.1535	150	6.6613	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1791	175	6.1911	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2047	200	5.9305	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2303	225	5.6825	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2559	250	5.5326	2.8771	0.2867	0.4619	0.7333	0.2835	0.5549	0.4056	0.2281	0.4883	0.9137	0.2555	0.5114	0.5220	0.4298	0.4673
0.2815	275	5.1671	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3071	300	5.2006	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3327	325	5.0447	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3582	350	4.9647	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3838	375	4.8521	2.5709	0.2881	0.4577	0.7438	0.2909	0.5712	0.4093	0.2273	0.5141	0.9008	0.2668	0.5117	0.5253	0.4331	0.4723
0.4094	400	4.8423	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4350	425	4.7472	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4606	450	4.6527	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4862	475	4.61	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5118	500	4.5451	2.4136	0.2786	0.4464	0.7485	0.2961	0.5638	0.4368	0.2269	0.5125	0.8998	0.2680	0.4938	0.5341	0.4383	0.4726
0.5374	525	4.5357	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5629	550	4.481	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5885	575	4.4669	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6141	600	4.3886	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6397	625	4.2929	2.3091	0.2639	0.4475	0.7521	0.3095	0.5619	0.4448	0.2244	0.5178	0.9102	0.2655	0.4809	0.5253	0.4351	0.4722
0.6653	650	4.2558	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6909	675	4.3228	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7165	700	4.2496	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7421	725	4.2304	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7677	750	4.224	2.2440	0.2628	0.4514	0.7387	0.3028	0.5522	0.4313	0.2253	0.5266	0.9211	0.2675	0.4929	0.5232	0.4351	0.4716
0.7932	775	4.2821	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8188	800	4.2686	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8444	825	4.1657	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8700	850	4.2297	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8956	875	4.1709	2.2142	0.2685	0.4520	0.7569	0.2930	0.5625	0.4486	0.2229	0.5280	0.9153	0.2601	0.4862	0.5199	0.4334	0.4729
0.9212	900	4.0771	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9468	925	4.1492	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9724	950	4.2074	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9980	975	4.0993	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0	977	-	-	0.2613	0.4507	0.7556	0.2964	0.5584	0.4416	0.2241	0.5271	0.9154	0.2646	0.4714	0.5211	0.4291	0.4705

Environmental Impact

Carbon emissions were measured using CodeCarbon.

• Energy Consumed: 0.643 kWh
• Carbon Emitted: 0.250 kg of CO2
• Hours Used: 1.727 hours

Training Hardware

• On Cloud: No
• GPU Model: 1 x NVIDIA GeForce RTX 3090
• CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
• RAM Size: 31.78 GB

Framework Versions

• Python: 3.11.6
• Sentence Transformers: 3.4.0.dev0
• Transformers: 4.46.2
• PyTorch: 2.5.0+cu121
• Accelerate: 0.35.0.dev0
• Datasets: 2.20.0
• Tokenizers: 0.20.3

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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

CachedMultipleNegativesRankingLoss

@misc{gao2021scaling,
    title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
    author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
    year={2021},
    eprint={2101.06983},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}