Bea-Taylor/objection_fine_tuned_4

SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2. It maps sentences & paragraphs to a 384-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: sentence-transformers/all-MiniLM-L6-v2
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity

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': 256, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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})
  (2): Normalize()
)

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("Bea-Taylor/objection_fine_tuned_4")
# Run inference
sentences = [
    'It will also reduce/limit/block east facing views of Canary Wharf for Swedish Quays residents next door. In turn, this will have a knock on effect on the value of our property because views of Canary Wharf are sought after by potential buyers.',
    'I support the planning application for the proposed development of the roof space. This project is a vital step toward easing the financial burden on residents and addressing ongoing concerns effectively. Additionally, it brings the added benefit of a positive environmental impact, contributing to a more sustainable and responsible community.',
    'My health has really suffered over the last two years during which, I have had 3 heart attacks and have recently been diagnosed with anaemia and emphysema as well as having stents surgically placed in my arteries. I am on twelve tablets a day for my health and this whole subject is creating all my health conditions to worsen. All I can do is emphasise my objections and hope that Barnet Council decline building permission.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

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

Evaluation

Metrics

Semantic Similarity

• Dataset: sts-dev
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.9829
spearman_cosine	0.9159

Training Details

Training Dataset

Unnamed Dataset

• Size: 180,000 training samples
• Columns: text1, text2, and label
• Approximate statistics based on the first 1000 samples:

  	text1	text2	label
  type	string	string	float
  details	min: 3 tokens mean: 50.78 tokens max: 256 tokens	min: 3 tokens mean: 50.72 tokens max: 256 tokens	min: 0.0 mean: 0.36 max: 1.0

• Samples:

  text1	text2	label
  Loss of Daylight and Sunlight	Fifthly, the increased height of the building with an additional storey will further reduce the available sunlight hours in my first floor flat leading to increased heating costs as well as a reduction in my quality of life. As the sun barely gets above the level of the existing building in the winter months it is likely I will spend much of the winter with the rear of my flat continually in shadow.	0.75
  The existing carpark has a maximum of 16 car parking spaces for the 37 flats, used on a first come first served basis.	As the other comments on this application state, the building works appear to be complete and the application does not appear to demonstrate the scale of the work or the reality of the build.	0.0
  Are you proposing to connect to the existing drainage system?	The design of the types of buildings being proposed is out of character with the area. I object to the removal of the existing footbridge... this is unacceptable and will cut the area of Victoria Park off from Cromer road - a currently safe route which pedestrians and school children use to access without having to cross road and the promise of new access or pedestrian routes will mean walking public pavements around New Barnet via Station Road. I am not convinced by the developers that an alternative safer route will be provided as they will say anything to obtain planning.	0.0

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

Evaluation Dataset

Unnamed Dataset

• Size: 20,000 evaluation samples
• Columns: text1, text2, and label
• Approximate statistics based on the first 1000 samples:

  	text1	text2	label
  type	string	string	float
  details	min: 3 tokens mean: 49.71 tokens max: 256 tokens	min: 4 tokens mean: 51.7 tokens max: 256 tokens	min: 0.0 mean: 0.33 max: 1.0

• Samples:

  text1	text2	label
  1. Significant noise and disruption for local residents of rainbow quay and surrounding developments - Including over looking Princes Court and blocking already limited light.	I object to the developer changing the goal posts in order to achieve more profit as it is not the families who need social housing who will benefit or the young people trying to get on the property ladder and can't but the developer who profits from those who can "afford" to pay the "high prices" of Barnet accommodation.	0.0
  Congestion on Camlet way and beach hill , roads which already have traffic issues!	TRAFFIC AND PARKING - Granville Road has limited off road parking and is a busy & important thoroughfare from Ballards Lane, to High Road North Finchley & Summers Lane, linking Finchley & Friern Barnet. The addition of two further flats without parking provision would increase the pressure for parking spaces.	0.75
  It will also obstruct light to my property and garden.	- Health and safety: concern for disruption building works will cause, damage to local infrastructure, increased traffic	0.0

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

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: 1
• warmup_ratio: 0.1
• fp16: 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: 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: 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: 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
• 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	sts-dev_spearman_cosine
-1	-1	-	-	0.3501
0.0089	100	0.1082	0.1113	0.3928
0.0178	200	0.1088	0.1000	0.4880
0.0267	300	0.0931	0.0857	0.5950
0.0356	400	0.0765	0.0749	0.6603
0.0444	500	0.0725	0.0685	0.6929
0.0533	600	0.066	0.0622	0.7216
0.0622	700	0.0568	0.0558	0.7506
0.0711	800	0.0525	0.0498	0.7749
0.08	900	0.048	0.0453	0.7926
0.0889	1000	0.0438	0.0412	0.8091
0.0978	1100	0.0447	0.0375	0.8239
0.1067	1200	0.0391	0.0336	0.8357
0.1156	1300	0.0359	0.0305	0.8486
0.1244	1400	0.0307	0.0272	0.8565
0.1333	1500	0.0289	0.0251	0.8621
0.1422	1600	0.0256	0.0233	0.8667
0.1511	1700	0.0285	0.0221	0.8702
0.16	1800	0.0229	0.0206	0.8743
0.1689	1900	0.0228	0.0193	0.8781
0.1778	2000	0.022	0.0182	0.8814
0.1867	2100	0.0197	0.0173	0.8827
0.1956	2200	0.0185	0.0167	0.8837
0.2044	2300	0.0202	0.0162	0.8850
0.2133	2400	0.0178	0.0151	0.8883
0.2222	2500	0.0174	0.0147	0.8896
0.2311	2600	0.0171	0.0146	0.8891
0.24	2700	0.0163	0.0136	0.8921
0.2489	2800	0.0147	0.0131	0.8934
0.2578	2900	0.0149	0.0129	0.8953
0.2667	3000	0.0152	0.0122	0.8966
0.2756	3100	0.0138	0.0120	0.8969
0.2844	3200	0.0128	0.0114	0.8977
0.2933	3300	0.0128	0.0111	0.8991
0.3022	3400	0.0117	0.0106	0.9005
0.3111	3500	0.0126	0.0104	0.9009
0.32	3600	0.0118	0.0102	0.9020
0.3289	3700	0.0115	0.0100	0.9016
0.3378	3800	0.0115	0.0098	0.9019
0.3467	3900	0.0116	0.0092	0.9035
0.3556	4000	0.0113	0.0090	0.9042
0.3644	4100	0.0117	0.0090	0.9043
0.3733	4200	0.0097	0.0084	0.9054
0.3822	4300	0.0098	0.0087	0.9052
0.3911	4400	0.0098	0.0085	0.9054
0.4	4500	0.0097	0.0084	0.9056
0.4089	4600	0.0097	0.0082	0.9057
0.4178	4700	0.0102	0.0080	0.9066
0.4267	4800	0.0086	0.0079	0.9071
0.4356	4900	0.0085	0.0078	0.9070
0.4444	5000	0.009	0.0076	0.9080
0.4533	5100	0.0091	0.0073	0.9085
0.4622	5200	0.0084	0.0073	0.9085
0.4711	5300	0.0082	0.0071	0.9089
0.48	5400	0.0073	0.0070	0.9089
0.4889	5500	0.0096	0.0069	0.9098
0.4978	5600	0.007	0.0068	0.9097
0.5067	5700	0.0078	0.0070	0.9096
0.5156	5800	0.0079	0.0067	0.9102
0.5244	5900	0.0097	0.0067	0.9107
0.5333	6000	0.0077	0.0065	0.9110
0.5422	6100	0.0084	0.0065	0.9112
0.5511	6200	0.007	0.0063	0.9113
0.56	6300	0.0073	0.0062	0.9117
0.5689	6400	0.0078	0.0066	0.9107
0.5778	6500	0.0082	0.0062	0.9116
0.5867	6600	0.0066	0.0061	0.9119
0.5956	6700	0.0076	0.0060	0.9122
0.6044	6800	0.0076	0.0060	0.9120
0.6133	6900	0.0075	0.0059	0.9123
0.6222	7000	0.0071	0.0059	0.9126
0.6311	7100	0.0076	0.0057	0.9130
0.64	7200	0.0067	0.0056	0.9131
0.6489	7300	0.0069	0.0057	0.9130
0.6578	7400	0.0068	0.0055	0.9134
0.6667	7500	0.0073	0.0054	0.9136
0.6756	7600	0.0063	0.0056	0.9131
0.6844	7700	0.0068	0.0054	0.9134
0.6933	7800	0.0057	0.0054	0.9135
0.7022	7900	0.0073	0.0053	0.9137
0.7111	8000	0.0063	0.0053	0.9139
0.72	8100	0.0061	0.0052	0.9139
0.7289	8200	0.0062	0.0052	0.9141
0.7378	8300	0.0065	0.0051	0.9143
0.7467	8400	0.0061	0.0052	0.9141
0.7556	8500	0.0064	0.0050	0.9146
0.7644	8600	0.0056	0.0050	0.9146
0.7733	8700	0.006	0.0050	0.9146
0.7822	8800	0.0066	0.0049	0.9147
0.7911	8900	0.005	0.0048	0.9150
0.8	9000	0.0056	0.0048	0.9149
0.8089	9100	0.0061	0.0048	0.9149
0.8178	9200	0.0057	0.0047	0.9149
0.8267	9300	0.0075	0.0048	0.9150
0.8356	9400	0.0057	0.0047	0.9152
0.8444	9500	0.0055	0.0047	0.9151
0.8533	9600	0.0056	0.0047	0.9153
0.8622	9700	0.0049	0.0047	0.9153
0.8711	9800	0.0066	0.0047	0.9154
0.88	9900	0.0054	0.0046	0.9154
0.8889	10000	0.0055	0.0046	0.9154
0.8978	10100	0.0055	0.0046	0.9155
0.9067	10200	0.0048	0.0045	0.9155
0.9156	10300	0.0046	0.0045	0.9156
0.9244	10400	0.0063	0.0045	0.9157
0.9333	10500	0.0055	0.0045	0.9157
0.9422	10600	0.0059	0.0045	0.9158
0.9511	10700	0.0049	0.0045	0.9158
0.96	10800	0.0058	0.0045	0.9158
0.9689	10900	0.0052	0.0045	0.9158
0.9778	11000	0.0065	0.0044	0.9159
0.9867	11100	0.0053	0.0044	0.9159
0.9956	11200	0.0046	0.0044	0.9159
-1	-1	-	-	0.9159

Framework Versions

• Python: 3.10.17
• Sentence Transformers: 4.1.0
• Transformers: 4.52.4
• PyTorch: 2.7.0
• 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",
}