Mossie96/all-mpnet-base-v2_distilled_3_layers_1-5-10

SentenceTransformer based on sentence-transformers/all-mpnet-base-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-mpnet-base-v2. 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: sentence-transformers/all-mpnet-base-v2
• Maximum Sequence Length: 384 tokens
• Output Dimensionality: 768 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': 384, 'do_lower_case': False}) with Transformer model: MPNetModel 
  (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})
  (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("sentence_transformers_model_id")
# Run inference
sentences = [
    'A black dog is drinking next to a brown and white dog that is looking at an orange ball in the lake, whilst a horse and rider passes behind.',
    'There are two people running around a track in lane three and the one wearing a blue shirt with a green thing over the eyes is just barely ahead of the guy wearing an orange shirt and sunglasses.',
    'the guy is dead',
]
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

Semantic Similarity

• Datasets: sts-dev and sts-test
• Evaluated with EmbeddingSimilarityEvaluator

Metric	sts-dev	sts-test
pearson_cosine	0.8659	0.8309
spearman_cosine	0.8685	0.8339

Knowledge Distillation

• Evaluated with MSEEvaluator

Metric	Value
negative_mse	-0.0158

Training Details

Training Dataset

Unnamed Dataset

• Size: 9,014,210 training samples
• Columns: sentence and label
• Approximate statistics based on the first 1000 samples:

  	sentence	label
  type	string	list
  details	min: 4 tokens mean: 12.24 tokens max: 52 tokens	size: 768 elements

• Samples:

  sentence	label
  A person on a horse jumps over a broken down airplane.	[-0.030610017478466034, 0.11742044985294342, 0.031586047261953354, 0.01859636977314949, 0.016319412738084793, ...]
  Children smiling and waving at camera	[-0.006198188289999962, -0.036625951528549194, -0.005352460313588381, -0.006725294981151819, 0.05185901001095772, ...]
  A boy is jumping on skateboard in the middle of a red bridge.	[-0.01783316768705845, -0.05204763263463974, -0.003716366598382592, 0.0009472182719036937, 0.05223219841718674, ...]

• Loss: MSELoss

Evaluation Dataset

Unnamed Dataset

• Size: 10,000 evaluation samples
• Columns: sentence and label
• Approximate statistics based on the first 1000 samples:

  	sentence	label
  type	string	list
  details	min: 5 tokens mean: 13.23 tokens max: 57 tokens	size: 768 elements

• Samples:

  sentence	label
  Two women are embracing while holding to go packages.	[0.010130808688700199, 0.009573593735694885, -0.00034817546838894486, -0.0040625291876494884, 0.02026110142469406, ...]
  Two young children in blue jerseys, one with the number 9 and one with the number 2 are standing on wooden steps in a bathroom and washing their hands in a sink.	[-0.033891696482896805, -0.04130887985229492, -0.006042165216058493, -0.02770376019179821, -0.0017171527724713087, ...]
  A man selling donuts to a customer during a world exhibition event held in the city of Angeles	[0.0013940087519586086, -0.044612932950258255, -0.023834265768527985, 0.11863800883293152, -0.03907289728522301, ...]

• Loss: MSELoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 64
• per_device_eval_batch_size: 64
• learning_rate: 0.0001
• num_train_epochs: 1
• warmup_ratio: 0.1
• fp16: True
• load_best_model_at_end: 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: 64
• per_device_eval_batch_size: 64
• 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: 0.0001
• 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: 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}
• tp_size: 0
• 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	negative_mse	sts-test_spearman_cosine
-1	-1	-	-	0.6786	-0.2176	-
0.0071	1000	0.0016	-	-	-	-
0.0142	2000	0.001	-	-	-	-
0.0213	3000	0.0008	-	-	-	-
0.0284	4000	0.0007	-	-	-	-
0.0355	5000	0.0006	0.0006	0.8511	-0.0561	-
0.0426	6000	0.0006	-	-	-	-
0.0497	7000	0.0005	-	-	-	-
0.0568	8000	0.0005	-	-	-	-
0.0639	9000	0.0005	-	-	-	-
0.0710	10000	0.0004	0.0004	0.8624	-0.0361	-
0.0781	11000	0.0004	-	-	-	-
0.0852	12000	0.0004	-	-	-	-
0.0923	13000	0.0004	-	-	-	-
0.0994	14000	0.0004	-	-	-	-
0.1065	15000	0.0003	0.0003	0.8649	-0.0288	-
0.1136	16000	0.0003	-	-	-	-
0.1207	17000	0.0003	-	-	-	-
0.1278	18000	0.0003	-	-	-	-
0.1349	19000	0.0003	-	-	-	-
0.1420	20000	0.0003	0.0003	0.8663	-0.0252	-
0.1491	21000	0.0003	-	-	-	-
0.1562	22000	0.0003	-	-	-	-
0.1633	23000	0.0003	-	-	-	-
0.1704	24000	0.0003	-	-	-	-
0.1775	25000	0.0003	0.0002	0.8641	-0.0232	-
0.1846	26000	0.0003	-	-	-	-
0.1917	27000	0.0003	-	-	-	-
0.1988	28000	0.0003	-	-	-	-
0.2059	29000	0.0003	-	-	-	-
0.2130	30000	0.0003	0.0002	0.8641	-0.0219	-
0.2201	31000	0.0003	-	-	-	-
0.2272	32000	0.0003	-	-	-	-
0.2343	33000	0.0003	-	-	-	-
0.2414	34000	0.0003	-	-	-	-
0.2485	35000	0.0003	0.0002	0.8649	-0.0209	-
0.2556	36000	0.0003	-	-	-	-
0.2627	37000	0.0003	-	-	-	-
0.2698	38000	0.0003	-	-	-	-
0.2769	39000	0.0003	-	-	-	-
0.2840	40000	0.0003	0.0002	0.8648	-0.0202	-
0.2911	41000	0.0003	-	-	-	-
0.2982	42000	0.0002	-	-	-	-
0.3053	43000	0.0002	-	-	-	-
0.3124	44000	0.0002	-	-	-	-
0.3195	45000	0.0002	0.0002	0.8663	-0.0196	-
0.3266	46000	0.0002	-	-	-	-
0.3337	47000	0.0002	-	-	-	-
0.3408	48000	0.0002	-	-	-	-
0.3479	49000	0.0002	-	-	-	-
0.3550	50000	0.0002	0.0002	0.8665	-0.0192	-
0.3621	51000	0.0002	-	-	-	-
0.3692	52000	0.0002	-	-	-	-
0.3763	53000	0.0002	-	-	-	-
0.3834	54000	0.0002	-	-	-	-
0.3905	55000	0.0002	0.0002	0.8650	-0.0187	-
0.3976	56000	0.0002	-	-	-	-
0.4047	57000	0.0002	-	-	-	-
0.4118	58000	0.0002	-	-	-	-
0.4189	59000	0.0002	-	-	-	-
0.4260	60000	0.0002	0.0002	0.8636	-0.0184	-
0.4331	61000	0.0002	-	-	-	-
0.4402	62000	0.0002	-	-	-	-
0.4473	63000	0.0002	-	-	-	-
0.4544	64000	0.0002	-	-	-	-
0.4615	65000	0.0002	0.0002	0.8673	-0.0180	-
0.4686	66000	0.0002	-	-	-	-
0.4757	67000	0.0002	-	-	-	-
0.4828	68000	0.0002	-	-	-	-
0.4899	69000	0.0002	-	-	-	-
0.4970	70000	0.0002	0.0002	0.8692	-0.0178	-
0.5041	71000	0.0002	-	-	-	-
0.5112	72000	0.0002	-	-	-	-
0.5183	73000	0.0002	-	-	-	-
0.5254	74000	0.0002	-	-	-	-
0.5325	75000	0.0002	0.0002	0.8675	-0.0175	-
0.5396	76000	0.0002	-	-	-	-
0.5467	77000	0.0002	-	-	-	-
0.5538	78000	0.0002	-	-	-	-
0.5609	79000	0.0002	-	-	-	-
0.5680	80000	0.0002	0.0002	0.8657	-0.0173	-
0.5751	81000	0.0002	-	-	-	-
0.5822	82000	0.0002	-	-	-	-
0.5893	83000	0.0002	-	-	-	-
0.5964	84000	0.0002	-	-	-	-
0.6035	85000	0.0002	0.0002	0.8670	-0.0171	-
0.6106	86000	0.0002	-	-	-	-
0.6177	87000	0.0002	-	-	-	-
0.6248	88000	0.0002	-	-	-	-
0.6319	89000	0.0002	-	-	-	-
0.6390	90000	0.0002	0.0002	0.8665	-0.0169	-
0.6461	91000	0.0002	-	-	-	-
0.6532	92000	0.0002	-	-	-	-
0.6603	93000	0.0002	-	-	-	-
0.6674	94000	0.0002	-	-	-	-
0.6745	95000	0.0002	0.0002	0.8672	-0.0167	-
0.6816	96000	0.0002	-	-	-	-
0.6887	97000	0.0002	-	-	-	-
0.6958	98000	0.0002	-	-	-	-
0.7029	99000	0.0002	-	-	-	-
0.7100	100000	0.0002	0.0002	0.8657	-0.0165	-
0.7171	101000	0.0002	-	-	-	-
0.7242	102000	0.0002	-	-	-	-
0.7313	103000	0.0002	-	-	-	-
0.7384	104000	0.0002	-	-	-	-
0.7455	105000	0.0002	0.0002	0.8676	-0.0165	-
0.7526	106000	0.0002	-	-	-	-
0.7597	107000	0.0002	-	-	-	-
0.7668	108000	0.0002	-	-	-	-
0.7739	109000	0.0002	-	-	-	-
0.7810	110000	0.0002	0.0002	0.8672	-0.0164	-
0.7881	111000	0.0002	-	-	-	-
0.7952	112000	0.0002	-	-	-	-
0.8023	113000	0.0002	-	-	-	-
0.8094	114000	0.0002	-	-	-	-
0.8165	115000	0.0002	0.0002	0.8698	-0.0162	-
0.8236	116000	0.0002	-	-	-	-
0.8307	117000	0.0002	-	-	-	-
0.8378	118000	0.0002	-	-	-	-
0.8449	119000	0.0002	-	-	-	-
0.8520	120000	0.0002	0.0002	0.8685	-0.0161	-
0.8591	121000	0.0002	-	-	-	-
0.8662	122000	0.0002	-	-	-	-
0.8733	123000	0.0002	-	-	-	-
0.8804	124000	0.0002	-	-	-	-
0.8875	125000	0.0002	0.0002	0.8676	-0.0160	-
0.8946	126000	0.0002	-	-	-	-
0.9017	127000	0.0002	-	-	-	-
0.9088	128000	0.0002	-	-	-	-
0.9159	129000	0.0002	-	-	-	-
0.9230	130000	0.0002	0.0002	0.8682	-0.0159	-
0.9301	131000	0.0002	-	-	-	-
0.9372	132000	0.0002	-	-	-	-
0.9443	133000	0.0002	-	-	-	-
0.9514	134000	0.0002	-	-	-	-
0.9585	135000	0.0002	0.0002	0.8678	-0.0158	-
0.9656	136000	0.0002	-	-	-	-
0.9727	137000	0.0002	-	-	-	-
0.9798	138000	0.0002	-	-	-	-
0.9869	139000	0.0002	-	-	-	-
0.9940	140000	0.0002	0.0002	0.8685	-0.0158	-
-1	-1	-	-	-	-	0.8339

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.11.11
• Sentence Transformers: 4.1.0
• Transformers: 4.51.3
• PyTorch: 2.7.1+cu118
• Accelerate: 1.7.0
• Datasets: 3.3.2
• 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",
}

MSELoss

@inproceedings{reimers-2020-multilingual-sentence-bert,
    title = "Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2020",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/2004.09813",
}