Tien09/tiny_bert_ft_sim_score_1231_1

MPNet base trained on AllNLI triplets

This is a sentence-transformers model finetuned from prajjwal1/bert-tiny on the pair_similarity_new_1231 dataset. It maps sentences & paragraphs to a 128-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: prajjwal1/bert-tiny
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 128 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • pair_similarity_new_1231
• 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': 128, '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("Tien09/tiny_bert_ft_sim_score_1231_1")
# Run inference
sentences = [
    'This card cannot be Normal Summoned or Set. This card cannot be Special Summoned except by the effect of "The First Sarcophagus". When this card is Special Summoned, you can Special Summon up to 4 Level 2 or lower Zombie-Type Normal Monsters from your GY.',
    'Target 1 Level 4 or lower "Magistus" monster in your GY; Special Summon it. If a "Magistus" card(s) in your Spell & Trap Zone would be destroyed by your opponent\'s card effect, you can banish this card from your GY instead. You can only use each effect of "Magistus Vritra" once per turn.',
    "When this card is Normal Summoned, you can remove from play 1 Psychic-Type monster from your Deck. This card's Level becomes the Level of that monster.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 128]

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

Training Details

Training Dataset

pair_similarity_new_1231

• Dataset: pair_similarity_new_1231 at 757b53a
• Size: 8,959 training samples
• Columns: effect_text, score, and effect_text2
• Approximate statistics based on the first 1000 samples:

  	effect_text	score	effect_text2
  type	string	float	string
  details	min: 9 tokens mean: 73.57 tokens max: 204 tokens	min: 0.0 mean: 0.42 max: 1.0	min: 8 tokens mean: 73.28 tokens max: 181 tokens

• Samples:

  effect_text	score	effect_text2
  When your opponent's monster attacks a face-up Level 4 or lower Toon Monster on your side of the field, you can make the attack a direct attack to your Life Points.	0.0	Cannot be used as material for a Fusion, Synchro, or Xyz Summon. Cannot be Tributed while face-up in the Monster Zone. If this card is sent to the GY as material for a Link Summon: Special Summon this card in Defense Position, to the opponent's field of the player that Link Summoned. You can only use this effect of "Click & Echo" twice per turn. While this card, that was Summoned by its effect, is in the Monster Zone, you must keep your hand revealed.
  When your opponent Special Summons a monster, you can discard 1 card to Special Summon this card from your hand. Your opponent cannot remove cards from play.	0.0	Once per turn you can place 2 Venom Counters on 1 monster your opponent controls. If you activate this effect, this card cannot attack during this turn.
  Mystical Elf" + "Curtain of the Dark Ones	0.0	If you Normal or Special Summon a "U.A." monster(s) (except during the Damage Step): You can Special Summon this card from your hand. If this card is Special Summoned: You can activate 1 of these effects.
  ● Target 1 card on the field; destroy it.
  ● Negate the effects of all face-up monsters on the field until the end of this turn, except "U.A." monsters.
  You can only use each effect of "U.A. Player Manager" once per turn.

• Loss: CoSENTLoss with these parameters:

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

Evaluation Dataset

pair_similarity_new_1231

• Dataset: pair_similarity_new_1231 at 757b53a
• Size: 1,920 evaluation samples
• Columns: effect_text, score, and effect_text2
• Approximate statistics based on the first 1000 samples:

  	effect_text	score	effect_text2
  type	string	float	string
  details	min: 6 tokens mean: 72.29 tokens max: 190 tokens	min: 0.0 mean: 0.42 max: 1.0	min: 5 tokens mean: 70.39 tokens max: 219 tokens

• Samples:

  effect_text	score	effect_text2
  2+ Level 4 monsters This Xyz Summoned card gains 500 ATK x the total Link Rating of Link Monsters linked to this card. You can detach 2 materials from this card, then target 1 4 Cyberse Link Monster in your GY; Special Summon it to your field so it points to this card, also you cannot Special Summon other monsters or attack directly for the rest of this turn.	1.0	3 Level 4 monsters Once per turn, you can also Xyz Summon "Zoodiac Tigermortar" by using 1 "Zoodiac" monster you control with a different name as Xyz Material. (If you used an Xyz Monster, any Xyz Materials attached to it also become Xyz Materials on this card.) This card gains ATK and DEF equal to the ATK and DEF of all "Zoodiac" monsters attached to it as Materials. Once per turn: You can detach 1 Xyz Material from this card, then target 1 Xyz Monster you control and 1 "Zoodiac" monster in your GY; attach that "Zoodiac" monster to that Xyz Monster as Xyz Material.
  1 Tuner + 1 or more non-Tuner Pendulum Monsters Once per turn: You can target 1 Pendulum Monster on the field or 1 card in the Pendulum Zone; destroy it, and if you do, shuffle 1 card on the field into the Deck. Once per turn: You can Special Summon 1 "Dracoslayer" monster from your Deck in Defense Position, but it cannot be used as a Synchro Material for a Summon.	0.5	You can Ritual Summon this card with a "Recipe" card. If this card is Special Summoned: You can target 1 Spell/Trap on the field; destroy it. When a card or effect is activated that targets this card on the field, or when this card is targeted for an attack (Quick Effect): You can Tribute this card and 1 Attack Position monster on either field, and if you do, Special Summon 1 Level 3 or 4 "Nouvelles" Ritual Monster from your hand or Deck. You can only use each effect of "Confiras de Nouvelles" once per turn.
  If you control an Illusion or Spellcaster monster: Add 1 "White Forest" monster from your Deck to your hand. If this card is sent to the GY to activate a monster effect: You can Set this card. You can only use each effect of "Tales of the White Forest" once per turn.	0.0	Once per turn, when your opponent activates a Trap Card, you can destroy the Trap Card and inflict 800 damage to your opponent.

• Loss: CoSENTLoss with these parameters:

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

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: 15
• warmup_ratio: 0.1
• fp16: 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: 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: 15
• 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
• 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
0.1786	100	4.6161	4.4244
0.3571	200	4.5391	4.3765
0.5357	300	4.3857	4.3268
0.7143	400	4.4125	4.2893
0.8929	500	4.2914	4.2755
1.0714	600	4.3029	4.2674
1.25	700	4.3832	4.2538
1.4286	800	4.2629	4.2410
1.6071	900	4.2343	4.2204
1.7857	1000	4.3121	4.2025
1.9643	1100	4.1853	4.1866
2.1429	1200	4.257	4.1712
2.3214	1300	4.269	4.1560
2.5	1400	4.1065	4.1373
2.6786	1500	4.1499	4.1204
2.8571	1600	4.1191	4.1044
3.0357	1700	4.0988	4.0989
3.2143	1800	4.1788	4.0736
3.3929	1900	4.0597	4.0633
3.5714	2000	4.0105	4.0565
3.75	2100	4.1035	4.0299
3.9286	2200	3.963	4.0527
4.1071	2300	4.0127	4.0191
4.2857	2400	4.0932	3.9967
4.4643	2500	3.9348	3.9900
4.6429	2600	3.9643	3.9798
4.8214	2700	3.9502	3.9671
5.0	2800	3.8734	3.9682
5.1786	2900	3.9211	3.9837
5.3571	3000	3.9833	3.9463
5.5357	3100	3.805	3.9531
5.7143	3200	3.9045	3.9353
5.8929	3300	3.7978	3.9654
6.0714	3400	3.8802	3.9545
6.25	3500	3.9052	3.9242
6.4286	3600	3.8237	3.9042
6.6071	3700	3.7338	3.9315
6.7857	3800	3.855	3.9185
6.9643	3900	3.7611	3.9310
7.1429	4000	3.8459	3.9072
7.3214	4100	3.8968	3.8727
7.5	4200	3.6306	3.9094
7.6786	4300	3.7761	3.8921
7.8571	4400	3.728	3.8924
8.0357	4500	3.7182	3.8869
8.2143	4600	3.7695	3.9223
8.3929	4700	3.7255	3.8472
8.5714	4800	3.6354	3.8880
8.75	4900	3.7751	3.8574
8.9286	5000	3.646	3.8901
9.1071	5100	3.7268	3.8599
9.2857	5200	3.7616	3.8432
9.4643	5300	3.6173	3.8403
9.6429	5400	3.6365	3.8573
9.8214	5500	3.6667	3.8416
10.0	5600	3.6456	3.8467
10.1786	5700	3.6096	3.8817
10.3571	5800	3.7093	3.8397
10.5357	5900	3.4805	3.8649
10.7143	6000	3.6849	3.8437
10.8929	6100	3.57	3.8676
11.0714	6200	3.6915	3.8439
11.25	6300	3.6812	3.8451
11.4286	6400	3.5948	3.8374
11.6071	6500	3.5601	3.8342
11.7857	6600	3.6627	3.8348
11.9643	6700	3.5013	3.8493
12.1429	6800	3.6723	3.8404
12.3214	6900	3.6744	3.8312
12.5	7000	3.521	3.8233
12.6786	7100	3.5399	3.8336
12.8571	7200	3.5862	3.8304
13.0357	7300	3.5598	3.8357
13.2143	7400	3.6152	3.8446
13.3929	7500	3.6303	3.8178
13.5714	7600	3.4542	3.8314
13.75	7700	3.6197	3.8265
13.9286	7800	3.4931	3.8349
14.1071	7900	3.6109	3.8310
14.2857	8000	3.6087	3.8320
14.4643	8100	3.5136	3.8299
14.6429	8200	3.5176	3.8318
14.8214	8300	3.607	3.8286
15.0	8400	3.5206	3.8300

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.3.1
• Transformers: 4.47.1
• PyTorch: 2.5.1+cu121
• Accelerate: 1.2.1
• Datasets: 3.2.0
• 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",
}

CoSENTLoss

@online{kexuefm-8847,
    title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
    author={Su Jianlin},
    year={2022},
    month={Jan},
    url={https://kexue.fm/archives/8847},
}