zhenqingli/all-MiniLM-L6-v2-ft-home-depot-positive-only

SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2 on the homedepot-search-anchor-positive-title-only dataset. It maps sentences & paragraphs to a 384-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-MiniLM-L6-v2
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • homedepot-search-anchor-positive-title-only

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': 256, 'do_lower_case': False, 'architecture': 'BertModel'})
  (1): Pooling({'word_embedding_dimension': 384, '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("zhenqingli/all-MiniLM-L6-v2-ft-home-depot-positive-only")
# Run inference
sentences = [
    'canopy',
    'Sunjoy Mojave 8 ft. x 5 ft. Steel Fabric Grill Gazebo',
    'Creek Pecan Tree',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.3364, 0.1916],
#         [0.3364, 1.0000, 0.1525],
#         [0.1916, 0.1525, 1.0000]])

Evaluation

Metrics

Semantic Similarity

• Dataset: homedepot_eval
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.3941
spearman_cosine	0.3677

Training Details

Training Dataset

homedepot-search-anchor-positive-title-only

• Dataset: homedepot-search-anchor-positive-title-only at 72eb0c2
• Size: 66,660 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 3 tokens mean: 6.62 tokens max: 24 tokens	min: 5 tokens mean: 20.24 tokens max: 47 tokens

• Samples:

  sentence1	sentence2
  ridiing lawnmower	Cub Cadet XT1 Enduro Series LT 42 in. 18 HP Kohler Hydrostatic Gas Front-Engine Riding Mower
  weather vanes	Olympia 24 in. x 24 in. x 62 in. Composite Vinyl Cupola with Copper Roof and Weathervane
  ryobi 24v	Ryobi 24-Volt Lithium-Ion Cordless String Trimmer/Edger with 24 in. Hedge Trimmer and 2 Batteries

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

homedepot-search-anchor-positive-title-only

• Dataset: homedepot-search-anchor-positive-title-only at 72eb0c2
• Size: 7,407 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 3 tokens mean: 6.34 tokens max: 17 tokens	min: 5 tokens mean: 20.04 tokens max: 51 tokens

• Samples:

  sentence1	sentence2
  bathroom sinks, double	KOHLER Ceramic/Impression 61 in. Vitreous China Double Vanity Top with Basin in White Impressions
  desoldering vacum pump	Aven Desoldering Pump with High Impact Plastic
  steel shelving	Edsal 36 in. W x 18 in. D x 72 in. H Steel Commercial Shelving Unit

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• num_train_epochs: 10
• load_best_model_at_end: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: epoch
• prediction_loss_only: True
• per_device_train_batch_size: 8
• per_device_eval_batch_size: 8
• 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: 1.0
• num_train_epochs: 10
• 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: 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}
• parallelism_config: None
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch_fused
• 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
• hub_revision: None
• 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
• liger_kernel_config: None
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	homedepot_eval_spearman_cosine
0.0120	100	0.2465	-	-
0.0240	200	0.2688	-	-
0.0360	300	0.2444	-	-
0.0480	400	0.1832	-	-
0.0600	500	0.2517	-	-
0.0720	600	0.2352	-	-
0.0840	700	0.1968	-	-
0.0960	800	0.227	-	-
0.1080	900	0.2342	-	-
0.1200	1000	0.2433	-	-
0.1320	1100	0.1859	-	-
0.1440	1200	0.2542	-	-
0.1560	1300	0.2137	-	-
0.1680	1400	0.1866	-	-
0.1800	1500	0.1791	-	-
0.1920	1600	0.1776	-	-
0.2040	1700	0.1806	-	-
0.2160	1800	0.2258	-	-
0.2280	1900	0.1834	-	-
0.2400	2000	0.1499	-	-
0.2520	2100	0.1587	-	-
0.2640	2200	0.1828	-	-
0.2760	2300	0.1855	-	-
0.2880	2400	0.1919	-	-
0.3000	2500	0.1573	-	-
0.3120	2600	0.1777	-	-
0.3240	2700	0.1958	-	-
0.3360	2800	0.1663	-	-
0.3480	2900	0.1883	-	-
0.3600	3000	0.2215	-	-
0.3720	3100	0.1518	-	-
0.3840	3200	0.1392	-	-
0.3960	3300	0.1261	-	-
0.4080	3400	0.1867	-	-
0.4200	3500	0.163	-	-
0.4320	3600	0.1924	-	-
0.4440	3700	0.173	-	-
0.4560	3800	0.1699	-	-
0.4680	3900	0.1479	-	-
0.4800	4000	0.1457	-	-
0.4920	4100	0.2051	-	-
0.5040	4200	0.1768	-	-
0.5160	4300	0.1328	-	-
0.5280	4400	0.182	-	-
0.5400	4500	0.1587	-	-
0.5520	4600	0.2071	-	-
0.5640	4700	0.1627	-	-
0.5760	4800	0.1497	-	-
0.5880	4900	0.1473	-	-
0.6000	5000	0.164	-	-
0.6120	5100	0.1353	-	-
0.6240	5200	0.1554	-	-
0.6360	5300	0.1448	-	-
0.6480	5400	0.1313	-	-
0.6600	5500	0.1469	-	-
0.6720	5600	0.1168	-	-
0.6840	5700	0.136	-	-
0.6960	5800	0.1255	-	-
0.7080	5900	0.1428	-	-
0.7200	6000	0.1263	-	-
0.7320	6100	0.1477	-	-
0.7440	6200	0.1774	-	-
0.7560	6300	0.1409	-	-
0.7680	6400	0.1352	-	-
0.7800	6500	0.1508	-	-
0.7920	6600	0.1127	-	-
0.8040	6700	0.1422	-	-
0.8160	6800	0.1393	-	-
0.8280	6900	0.1276	-	-
0.8400	7000	0.1457	-	-
0.8520	7100	0.1195	-	-
0.8640	7200	0.1459	-	-
0.8760	7300	0.1041	-	-
0.8880	7400	0.1657	-	-
0.9000	7500	0.1331	-	-
0.9120	7600	0.1116	-	-
0.9240	7700	0.1166	-	-
0.9360	7800	0.1252	-	-
0.9480	7900	0.106	-	-
0.9600	8000	0.1201	-	-
0.9720	8100	0.0966	-	-
0.9840	8200	0.119	-	-
0.9960	8300	0.1433	-	-
1.0	8333	-	0.1188	0.3844
1.0080	8400	0.0692	-	-
1.0200	8500	0.0918	-	-
1.0320	8600	0.0851	-	-
1.0440	8700	0.0792	-	-
1.0560	8800	0.0843	-	-
1.0680	8900	0.0832	-	-
1.0800	9000	0.0682	-	-
1.0920	9100	0.0854	-	-
1.1040	9200	0.0819	-	-
1.1160	9300	0.0711	-	-
1.1280	9400	0.0886	-	-
1.1400	9500	0.1026	-	-
1.1520	9600	0.0764	-	-
1.1640	9700	0.076	-	-
1.1760	9800	0.0933	-	-
1.1880	9900	0.0828	-	-
1.2000	10000	0.1089	-	-
1.2120	10100	0.063	-	-
1.2240	10200	0.1106	-	-
1.2360	10300	0.075	-	-
1.2480	10400	0.0856	-	-
1.2601	10500	0.097	-	-
1.2721	10600	0.0653	-	-
1.2841	10700	0.0702	-	-
1.2961	10800	0.078	-	-
1.3081	10900	0.0901	-	-
1.3201	11000	0.0742	-	-
1.3321	11100	0.0863	-	-
1.3441	11200	0.0701	-	-
1.3561	11300	0.0946	-	-
1.3681	11400	0.0901	-	-
1.3801	11500	0.0774	-	-
1.3921	11600	0.0531	-	-
1.4041	11700	0.0747	-	-
1.4161	11800	0.0873	-	-
1.4281	11900	0.0755	-	-
1.4401	12000	0.0947	-	-
1.4521	12100	0.1054	-	-
1.4641	12200	0.0753	-	-
1.4761	12300	0.1014	-	-
1.4881	12400	0.0724	-	-
1.5001	12500	0.0722	-	-
1.5121	12600	0.071	-	-
1.5241	12700	0.0765	-	-
1.5361	12800	0.0731	-	-
1.5481	12900	0.0799	-	-
1.5601	13000	0.1066	-	-
1.5721	13100	0.0537	-	-
1.5841	13200	0.0978	-	-
1.5961	13300	0.0891	-	-
1.6081	13400	0.0587	-	-
1.6201	13500	0.0736	-	-
1.6321	13600	0.0926	-	-
1.6441	13700	0.0586	-	-
1.6561	13800	0.0603	-	-
1.6681	13900	0.0898	-	-
1.6801	14000	0.1111	-	-
1.6921	14100	0.0889	-	-
1.7041	14200	0.1008	-	-
1.7161	14300	0.1004	-	-
1.7281	14400	0.0845	-	-
1.7401	14500	0.0993	-	-
1.7521	14600	0.081	-	-
1.7641	14700	0.0906	-	-
1.7761	14800	0.1068	-	-
1.7881	14900	0.0715	-	-
1.8001	15000	0.0805	-	-
1.8121	15100	0.0807	-	-
1.8241	15200	0.093	-	-
1.8361	15300	0.0921	-	-
1.8481	15400	0.083	-	-
1.8601	15500	0.088	-	-
1.8721	15600	0.0932	-	-
1.8841	15700	0.0786	-	-
1.8961	15800	0.0653	-	-
1.9081	15900	0.0833	-	-
1.9201	16000	0.0989	-	-
1.9321	16100	0.0853	-	-
1.9441	16200	0.0656	-	-
1.9561	16300	0.0775	-	-
1.9681	16400	0.0605	-	-
1.9801	16500	0.0802	-	-
1.9921	16600	0.075	-	-
2.0	16666	-	0.0885	0.3778
2.0041	16700	0.0621	-	-
2.0161	16800	0.0554	-	-
2.0281	16900	0.0493	-	-
2.0401	17000	0.0805	-	-
2.0521	17100	0.0516	-	-
2.0641	17200	0.0503	-	-
2.0761	17300	0.0624	-	-
2.0881	17400	0.0599	-	-
2.1001	17500	0.0532	-	-
2.1121	17600	0.052	-	-
2.1241	17700	0.0469	-	-
2.1361	17800	0.0458	-	-
2.1481	17900	0.0399	-	-
2.1601	18000	0.0629	-	-
2.1721	18100	0.0716	-	-
2.1841	18200	0.06	-	-
2.1961	18300	0.0309	-	-
2.2081	18400	0.0365	-	-
2.2201	18500	0.0542	-	-
2.2321	18600	0.0584	-	-
2.2441	18700	0.0686	-	-
2.2561	18800	0.0473	-	-
2.2681	18900	0.0564	-	-
2.2801	19000	0.0431	-	-
2.2921	19100	0.0557	-	-
2.3041	19200	0.0414	-	-
2.3161	19300	0.0528	-	-
2.3281	19400	0.0578	-	-
2.3401	19500	0.0552	-	-
2.3521	19600	0.0546	-	-
2.3641	19700	0.0448	-	-
2.3761	19800	0.0656	-	-
2.3881	19900	0.0466	-	-
2.4001	20000	0.0498	-	-
2.4121	20100	0.0563	-	-
2.4241	20200	0.0384	-	-
2.4361	20300	0.0254	-	-
2.4481	20400	0.0461	-	-
2.4601	20500	0.0492	-	-
2.4721	20600	0.0514	-	-
2.4841	20700	0.0503	-	-
2.4961	20800	0.0649	-	-
2.5081	20900	0.0617	-	-
2.5201	21000	0.0446	-	-
2.5321	21100	0.055	-	-
2.5441	21200	0.0606	-	-
2.5561	21300	0.05	-	-
2.5681	21400	0.043	-	-
2.5801	21500	0.0585	-	-
2.5921	21600	0.0519	-	-
2.6041	21700	0.0717	-	-
2.6161	21800	0.0516	-	-
2.6281	21900	0.0631	-	-
2.6401	22000	0.0397	-	-
2.6521	22100	0.045	-	-
2.6641	22200	0.0412	-	-
2.6761	22300	0.0455	-	-
2.6881	22400	0.0556	-	-
2.7001	22500	0.0475	-	-
2.7121	22600	0.0428	-	-
2.7241	22700	0.053	-	-
2.7361	22800	0.0552	-	-
2.7481	22900	0.0542	-	-
2.7601	23000	0.039	-	-
2.7721	23100	0.0535	-	-
2.7841	23200	0.0558	-	-
2.7961	23300	0.0503	-	-
2.8081	23400	0.0371	-	-
2.8201	23500	0.0497	-	-
2.8321	23600	0.037	-	-
2.8441	23700	0.0598	-	-
2.8561	23800	0.0504	-	-
2.8681	23900	0.0501	-	-
2.8801	24000	0.0375	-	-
2.8921	24100	0.0425	-	-
2.9041	24200	0.0447	-	-
2.9161	24300	0.0408	-	-
2.9281	24400	0.0618	-	-
2.9401	24500	0.0504	-	-
2.9521	24600	0.0585	-	-
2.9641	24700	0.0432	-	-
2.9761	24800	0.0403	-	-
2.9881	24900	0.0483	-	-
3.0	24999	-	0.0845	0.3677

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.11.5
• Sentence Transformers: 5.1.0
• Transformers: 4.56.1
• PyTorch: 2.8.0+cu128
• Accelerate: 1.10.1
• Datasets: 4.0.0
• Tokenizers: 0.22.0

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}
}