am-azadi/UAE-Large-V1_Fine_Tuned_2e

SentenceTransformer based on am-azadi/UAE-Large-V1_Fine_Tuned

This is a sentence-transformers model finetuned from am-azadi/UAE-Large-V1_Fine_Tuned. It maps sentences & paragraphs to a 1024-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: am-azadi/UAE-Large-V1_Fine_Tuned
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 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': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Look what a show Pope Francis gave in yesterday\'s homily / sermon! It\'s to be read and reread over and over again... This is the most spiritual Pope since Peter. "You may have flaws, be anxious, and sometimes live irritated, but don\'t forget that your life is the greatest company in the world. Only you can prevent it from going into decline. Many appreciate you, admire you and love you. remember that being happy is not having a sky without storms, a road without accidents, work without fatigue, relationships without disappointments. Being happy is finding strength in forgiveness, hope in battles, security in the stage of fear, love in discord. It\'s not just appreciating the smile, but also reflecting on sadness. It\'s not just celebrating successes, but learning lessons from failures. It\'s not just feeling happy with applause, but being happy in anonymity. Being happy is recognizing that it\'s worth life is worth living, despite all the challenges, misunderstandings, periods of crisis. Being happy is not a fatality of fate, but an achievement for those who manage to travel within themselves. To be happy is to stop feeling like a victim of problems and become the author of his own story . It\'s crossing deserts outside of yourself, but managing to find an oasis in the depths of our soul. It is to thank God for each morning, for the miracle of life. Being happy is not being afraid of your own feelings. It\'s knowing how to talk about yourself. It\'s having the courage to hear a "no". It\'s feeling safe when receiving criticism, even if unfair. It\'s kissing the children, pampering the parents, living poetic moments with friends, even when they hurt us. To be happy is to let the creature that lives in each of us live, free, joyful and simple. It\'s having maturity to be able to say: "I was wrong". It\'s having the courage to say, "I\'m sorry". It\'s having the sensitivity to say: "I need you". It\'s having the ability to say, "I love you". May your life become a garden of opportunities to be happy... May your springtime be a lover of joy. May in your winters be a lover of wisdom. And that when you make a mistake, start over from the beginning. For only then will you be in love with life. You will discover that being happy is not having a perfect life. But using tears to irrigate tolerance. Use losses to train patience. Using mistakes to sculpt serenity. Using pain to cut pleasure. Use obstacles to open intelligence windows. Never give up....Never give up on the people who love you. Never give up happiness, for life is an incredible spectacle." (Pope Francis).',
    '"The message that Pope Francis gave in yesterday\'s homily/sermon! It is to be read and reread several times... What an admirable man!"',
    'Denmark allows Muslim women to wear the niqab',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

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

Training Details

Training Dataset

Unnamed Dataset

• Size: 25,743 training samples
• Columns: sentence_0, sentence_1, and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	label
  type	string	string	float
  details	min: 6 tokens mean: 113.55 tokens max: 512 tokens	min: 4 tokens mean: 17.89 tokens max: 126 tokens	min: 1.0 mean: 1.0 max: 1.0

• Samples:

  sentence_0	sentence_1	label
  de. has left the route with his the ultimate nazi prey dogs. With his patches with swastikas of Tito Adolfito Hitler and those little things. But don't tell them fucking Nazis, they'll be offended later 180 * Rov	Santiago Abascal posed next to a man wearing Nazi emblems at a biker rally in Valladolid	1.0
  the info is ... a Danish police officer told a woman who was wearing a veil that the parliament had decided to approve the use of the niqab for Muslim women in Denmark. The situation is reversed in Indonesia, which (he said) has a Muslim majority population. Aya naon with the country ieu?	Denmark allows Muslim women to wear the niqab	1.0
  In Kolwezi, a Congolese driver destroys Chinese trucks out of anger	A Congolese destroys the trucks of a Chinese company in the DRC	1.0

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 2
• per_device_eval_batch_size: 2
• num_train_epochs: 1
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: no
• prediction_loss_only: True
• per_device_train_batch_size: 2
• per_device_eval_batch_size: 2
• 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: 5e-05
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1
• num_train_epochs: 1
• 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: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: False
• 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: 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: round_robin

Training Logs

Epoch	Step	Training Loss
0.0388	500	0.0173
0.0777	1000	0.0124
0.1165	1500	0.0127
0.1554	2000	0.0256
0.1942	2500	0.0123
0.2331	3000	0.0199
0.2719	3500	0.0079
0.3108	4000	0.0134
0.3496	4500	0.0127
0.3884	5000	0.026
0.4273	5500	0.0314
0.4661	6000	0.0267
0.5050	6500	0.0145
0.5438	7000	0.0093
0.5827	7500	0.007
0.6215	8000	0.0071
0.6603	8500	0.0116
0.6992	9000	0.0085
0.7380	9500	0.0157
0.7769	10000	0.0051
0.8157	10500	0.0101
0.8546	11000	0.0174
0.8934	11500	0.0116
0.9323	12000	0.0073
0.9711	12500	0.0146

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 3.4.1
• Transformers: 4.48.3
• PyTorch: 2.5.1+cu124
• Accelerate: 1.3.0
• Datasets: 3.3.1
• Tokenizers: 0.21.0

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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}