jonny9f/food_embeddings

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("jonny9f/food_embeddings")
# Run inference
sentences = [
    'Beef Top Round, lean raw',
    'Luncheon Slices, meatless',
    'Pasta Sauce, spaghetti/marinara ready-to-serve',
]
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

• Dataset: validation
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.9913
spearman_cosine	0.9868

Training Details

Training Dataset

Unnamed Dataset

• Size: 1,200,000 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: 4 tokens mean: 10.2 tokens max: 28 tokens	min: 4 tokens mean: 9.65 tokens max: 23 tokens	min: 0.0 mean: 0.26 max: 0.92

• Samples:

  sentence_0	sentence_1	label
  Beef top round roast, boneless lean select cooked	Blueberries, canned wild in heavy syrup drained	0.21440656185150148
  Nance, frozen unsweetened	Soymilk, unsweetened	0.3654276132583618
  Drops - Lemonade	Pickle relish, sweet	0.30108280181884767

• Loss: CosineSimilarityLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• 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: 32
• per_device_eval_batch_size: 32
• 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

Click to expand

Epoch	Step	Training Loss	validation_spearman_cosine
0.0133	500	0.0031	-
0.0267	1000	0.0028	-
0.04	1500	0.0025	-
0.0533	2000	0.0024	-
0.0667	2500	0.0023	-
0.08	3000	0.0022	-
0.0933	3500	0.0021	-
0.1067	4000	0.002	-
0.12	4500	0.002	-
0.1333	5000	0.0019	-
0.1467	5500	0.0018	-
0.16	6000	0.0018	-
0.1733	6500	0.0017	-
0.1867	7000	0.0017	-
0.2	7500	0.0016	-
0.2133	8000	0.0016	-
0.2267	8500	0.0016	-
0.24	9000	0.0015	-
0.2533	9500	0.0015	-
0.2667	10000	0.0015	-
0.28	10500	0.0015	-
0.2933	11000	0.0015	-
0.3067	11500	0.0014	-
0.32	12000	0.0014	-
0.3333	12500	0.0013	-
0.3467	13000	0.0013	-
0.36	13500	0.0013	-
0.3733	14000	0.0013	-
0.3867	14500	0.0012	-
0.4	15000	0.0012	-
0.4133	15500	0.0012	-
0.4267	16000	0.0012	-
0.44	16500	0.0012	-
0.4533	17000	0.0012	-
0.4667	17500	0.0011	-
0.48	18000	0.0011	-
0.4933	18500	0.0011	-
0.5067	19000	0.0011	-
0.52	19500	0.0011	-
0.5333	20000	0.0011	-
0.5467	20500	0.0011	-
0.56	21000	0.001	-
0.5733	21500	0.001	-
0.5867	22000	0.001	-
0.6	22500	0.001	-
0.6133	23000	0.001	-
0.6267	23500	0.001	-
0.64	24000	0.0009	-
0.6533	24500	0.0009	-
0.6667	25000	0.0009	-
0.68	25500	0.0009	-
0.6933	26000	0.0009	-
0.7067	26500	0.0009	-
0.72	27000	0.0009	-
0.7333	27500	0.0009	-
0.7467	28000	0.0009	-
0.76	28500	0.0008	-
0.7733	29000	0.0008	-
0.7867	29500	0.0008	-
0.8	30000	0.0008	-
0.8133	30500	0.0008	-
0.8267	31000	0.0008	-
0.84	31500	0.0008	-
0.8533	32000	0.0008	-
0.8667	32500	0.0008	-
0.88	33000	0.0007	-
0.8933	33500	0.0007	-
0.9067	34000	0.0008	-
0.92	34500	0.0007	-
0.9333	35000	0.0007	-
0.9467	35500	0.0007	-
0.96	36000	0.0007	-
0.9733	36500	0.0007	-
0.9867	37000	0.0007	-
1.0	37500	0.0007	0.9799
0.0133	500	0.0009	-
0.0267	1000	0.0011	-
0.04	1500	0.0011	-
0.0533	2000	0.001	-
0.0667	2500	0.001	-
0.08	3000	0.001	-
0.0933	3500	0.001	-
0.1067	4000	0.001	-
0.12	4500	0.001	-
0.1333	5000	0.001	-
0.1467	5500	0.001	-
0.16	6000	0.0009	-
0.1733	6500	0.0009	-
0.1867	7000	0.0009	-
0.2	7500	0.0009	-
0.2133	8000	0.001	-
0.2267	8500	0.0009	-
0.24	9000	0.0009	-
0.2533	9500	0.0009	-
0.2667	10000	0.0008	-
0.28	10500	0.0009	-
0.2933	11000	0.0008	-
0.3067	11500	0.0008	-
0.32	12000	0.0008	-
0.3333	12500	0.0008	-
0.3467	13000	0.0008	-
0.36	13500	0.0008	-
0.3733	14000	0.0008	-
0.3867	14500	0.0008	-
0.4	15000	0.0008	-
0.4133	15500	0.0007	-
0.4267	16000	0.0007	-
0.44	16500	0.0008	-
0.4533	17000	0.0007	-
0.4667	17500	0.0007	-
0.48	18000	0.0007	-
0.4933	18500	0.0007	-
0.5067	19000	0.0007	-
0.52	19500	0.0007	-
0.5333	20000	0.0007	-
0.5467	20500	0.0007	-
0.56	21000	0.0007	-
0.5733	21500	0.0006	-
0.5867	22000	0.0007	-
0.6	22500	0.0006	-
0.6133	23000	0.0006	-
0.6267	23500	0.0006	-
0.64	24000	0.0006	-
0.6533	24500	0.0006	-
0.6667	25000	0.0006	-
0.68	25500	0.0006	-
0.6933	26000	0.0006	-
0.7067	26500	0.0006	-
0.72	27000	0.0006	-
0.7333	27500	0.0006	-
0.7467	28000	0.0006	-
0.76	28500	0.0005	-
0.7733	29000	0.0005	-
0.7867	29500	0.0006	-
0.8	30000	0.0005	-
0.8133	30500	0.0005	-
0.8267	31000	0.0005	-
0.84	31500	0.0005	-
0.8533	32000	0.0005	-
0.8667	32500	0.0005	-
0.88	33000	0.0005	-
0.8933	33500	0.0005	-
0.9067	34000	0.0005	-
0.92	34500	0.0005	-
0.9333	35000	0.0005	-
0.9467	35500	0.0005	-
0.96	36000	0.0005	-
0.9733	36500	0.0005	-
0.9867	37000	0.0005	-
1.0	37500	0.0005	0.9850
0.0133	500	0.0004	-
0.0267	1000	0.0005	-
0.04	1500	0.0005	-
0.0533	2000	0.0005	-
0.0667	2500	0.0005	-
0.08	3000	0.0005	-
0.0933	3500	0.0005	-
0.1067	4000	0.0004	-
0.12	4500	0.0004	-
0.1333	5000	0.0004	-
0.1467	5500	0.0004	-
0.16	6000	0.0004	-
0.1733	6500	0.0004	-
0.1867	7000	0.0004	-
0.2	7500	0.0004	-
0.2133	8000	0.0004	-
0.2267	8500	0.0004	-
0.24	9000	0.0004	-
0.2533	9500	0.0004	-
0.2667	10000	0.0004	-
0.28	10500	0.0004	-
0.2933	11000	0.0004	-
0.3067	11500	0.0004	-
0.32	12000	0.0004	-
0.3333	12500	0.0004	-
0.3467	13000	0.0004	-
0.36	13500	0.0004	-
0.3733	14000	0.0004	-
0.3867	14500	0.0004	-
0.4	15000	0.0004	-
0.4133	15500	0.0004	-
0.4267	16000	0.0004	-
0.44	16500	0.0004	-
0.4533	17000	0.0004	-
0.4667	17500	0.0004	-
0.48	18000	0.0004	-
0.4933	18500	0.0004	-
0.5067	19000	0.0004	-
0.52	19500	0.0004	-
0.5333	20000	0.0004	-
0.5467	20500	0.0004	-
0.56	21000	0.0004	-
0.5733	21500	0.0004	-
0.5867	22000	0.0004	-
0.6	22500	0.0004	-
0.6133	23000	0.0004	-
0.6267	23500	0.0004	-
0.64	24000	0.0004	-
0.6533	24500	0.0004	-
0.6667	25000	0.0004	-
0.68	25500	0.0004	-
0.6933	26000	0.0004	-
0.7067	26500	0.0004	-
0.72	27000	0.0004	-
0.7333	27500	0.0004	-
0.7467	28000	0.0004	-
0.76	28500	0.0004	-
0.7733	29000	0.0004	-
0.7867	29500	0.0004	-
0.8	30000	0.0004	-
0.8133	30500	0.0004	-
0.8267	31000	0.0004	-
0.84	31500	0.0004	-
0.8533	32000	0.0004	-
0.8667	32500	0.0004	-
0.88	33000	0.0004	-
0.8933	33500	0.0004	-
0.9067	34000	0.0004	-
0.92	34500	0.0004	-
0.9333	35000	0.0004	-
0.9467	35500	0.0004	-
0.96	36000	0.0004	-
0.9733	36500	0.0004	-
0.9867	37000	0.0004	-
1.0	37500	0.0004	0.9868

Framework Versions

• Python: 3.11.3
• Sentence Transformers: 3.3.1
• Transformers: 4.48.0
• PyTorch: 2.5.1+cu124
• 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",
}