ebinan92/dunzhang_stella_en_1.5B_v5

SentenceTransformer based on dunzhang/stella_en_1.5B_v5

This is a sentence-transformers model finetuned from dunzhang/stella_en_1.5B_v5. 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: dunzhang/stella_en_1.5B_v5
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 tokens
• 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: Qwen2Model 
  (1): Pooling({'word_embedding_dimension': 1536, '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): Dense({'in_features': 1536, 'out_features': 1024, 'bias': True, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

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 = [
    'QuestionSummary: Function Machines\nQuestion: Which of the following pairs of function machines are correct?\nCorrectAnswer: $aRightarrowtimes2Rightarrow-5Rightarrow2a-5a\; \backslash \backslash Rightarrow\; \backslash \backslash times2\; \backslash \backslash Rightarrow\; -5\backslash \backslash Rightarrow\; 2a-5$ \n\n\\(a \\Rightarrow -5  \\Rightarrow \\times2\\Rightarrow 2(a-5)\\) \nAnswer: $aRightarrowtimes2Rightarrow-5Rightarrow2a-5a\; \backslash \backslash Rightarrow\; \backslash \backslash times2\; \backslash \backslash Rightarrow\; -5\backslash \backslash Rightarrow\; 2a-5$ \n\n\\(a \\Rightarrow \\times2  \\Rightarrow -5\\Rightarrow 2(a-5)\\) ',
    'Does not follow the arrows through a function machine, changes the order of the operations asked.',
    'Incorrectly cancels what they believe is a factor in algebraic fractions',
]
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]

Evaluation

Metrics

Information Retrieval

• Dataset: val
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@25	0.6946
cosine_precision@100	0.0088
cosine_precision@200	0.0047
cosine_precision@300	0.0032
cosine_precision@400	0.0024
cosine_precision@500	0.002
cosine_precision@600	0.0016
cosine_precision@700	0.0014
cosine_precision@800	0.0012
cosine_precision@900	0.0011
cosine_precision@1000	0.001
cosine_recall@100	0.876
cosine_recall@200	0.9369
cosine_recall@300	0.9587
cosine_recall@400	0.9736
cosine_recall@500	0.9805
cosine_recall@600	0.9862
cosine_recall@700	0.9931
cosine_recall@800	0.9954
cosine_recall@900	0.9977
cosine_recall@1000	0.9977
cosine_ndcg@25	0.3564
cosine_mrr@25	0.261
cosine_map@25	0.261
dot_accuracy@25	0.4271
dot_precision@100	0.0076
dot_precision@200	0.0043
dot_precision@300	0.0031
dot_precision@400	0.0024
dot_precision@500	0.0019
dot_precision@600	0.0016
dot_precision@700	0.0014
dot_precision@800	0.0012
dot_precision@900	0.0011
dot_precision@1000	0.001
dot_recall@100	0.76
dot_recall@200	0.8657
dot_recall@300	0.9231
dot_recall@400	0.9437
dot_recall@500	0.961
dot_recall@600	0.9713
dot_recall@700	0.9793
dot_recall@800	0.9839
dot_recall@900	0.9874
dot_recall@1000	0.9897
dot_ndcg@25	0.1953
dot_mrr@25	0.1329
dot_map@25	0.1329

Training Details

Training Dataset

Unnamed Dataset

• Size: 3,999 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 30 tokens mean: 87.03 tokens max: 363 tokens	min: 4 tokens mean: 13.84 tokens max: 42 tokens

• Samples:

  anchor	positive
  QuestionSummary: Simplifying Algebraic Fractions Question: Simplify the following, if possible: ( \frac{m^{2}+2 m-3}{m-3} ) CorrectAnswer: Does not simplify Answer: ( m+1 )	Does not know that to factorise a quadratic expression, to find two numbers that add to give the coefficient of the x term, and multiply to give the non variable term
  QuestionSummary: Range and Interquartile Range from a List of Data Question: Tom and Katie are discussing the ( 5 ) plants with these heights: ( 24 \mathrm{cm}, 17 \mathrm{cm}, 42 \mathrm{cm}, 26 \mathrm{cm}, 13 \mathrm{cm} ) Tom says if all the plants were cut in half, the range wouldn't change. Katie says if all the plants grew by ( 3 \mathrm{cm} ) each, the range wouldn't change. Who do you agree with? CorrectAnswer: Only Katie Answer: Only Tom	Believes if you changed all values by the same proportion the range would not change
  QuestionSummary: Properties of Quadrilaterals Question: The angles highlighted on this rectangle with different length sides can never be... CorrectAnswer: ( 90^{\circ} ) Answer: acute	Does not know the properties of a rectangle

• Loss: CachedMultipleNegativesSymmetricRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "mini_batch_size": 1
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 1372
• per_device_eval_batch_size: 1372
• learning_rate: 4e-05
• num_train_epochs: 5
• warmup_ratio: 0.1
• save_only_model: True
• bf16: True
• load_best_model_at_end: 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: 1372
• per_device_eval_batch_size: 1372
• 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: 4e-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: 5
• 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: True
• 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: 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: 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}
• 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
• 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
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	val_cosine_map@25
0.3333	1	2.2717	0.1775
0.6667	2	2.1785	0.2300
1.0	3	1.4112	0.2651
1.3333	4	1.1861	0.2726
1.6667	5	0.8742	0.2813
2.0	6	0.8327	0.2818
2.3333	7	0.7626	0.2777
2.6667	8	0.5767	0.2752
3.0	9	0.493	0.2698
3.3333	10	0.5174	0.2654
3.6667	11	0.3906	0.2655
4.0	12	0.419	0.2627
4.3333	13	0.4394	0.2625
4.6667	14	0.5449	0.2612
5.0	15	0.3731	0.2610

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.13
• Sentence Transformers: 3.1.1
• Transformers: 4.45.1
• PyTorch: 2.2.0
• Accelerate: 0.34.2
• Datasets: 3.0.1
• Tokenizers: 0.20.0

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",
}