SentenceTransformer based on sentence-transformers/multi-qa-MiniLM-L6-cos-v1

This is a sentence-transformers model finetuned from sentence-transformers/multi-qa-MiniLM-L6-cos-v1. 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/multi-qa-MiniLM-L6-cos-v1
Maximum Sequence Length: 512 tokens
Output Dimensionality: 384 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: 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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Must have experience in software development',
    "CEO of IMPROVLearning, specializing in e-learning and driver education. Founded and managed multiple ventures in training, healthcare, and real estate. Proven track record of expanding product offerings and achieving recognition on the Inc 500/5000 list. Active board member of the LA Chapter of the Entrepreneur Organization, contributing to the growth of over 3 million students. Experienced in venture capital and entrepreneurship, with a focus on innovative training solutions and community engagement. Active member of various organizations, including the Entrepreneurs' Organization and the Los Angeles County Business Federation.",
    'Multi-skilled Business Analytics professional with a Master’s in Business Analytics and a dual MBA. Experienced in data analytics, predictive modeling, and project management within the health and wellness sector. Proficient in extracting, summarizing, and analyzing claims databases and healthcare analytics. Skilled in statistical analysis, database management, and data visualization. Previous roles include Business Analytics Advisor at Cigna Healthcare and Informatics Senior Specialist at Cigna Healthcare. Strong leadership and project management abilities, with a solid foundation in healthcare economics and outcomes observational research. Familiar with Base SAS 9.2, SAS EG, SAS EM, SAS JMP, Tableau, and Oracle Crystal Ball.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# 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	0.9594
spearman_cosine	0.8606
pearson_manhattan	0.9437
spearman_manhattan	0.8604
pearson_euclidean	0.9443
spearman_euclidean	0.8606
pearson_dot	0.9594
spearman_dot	0.8606
pearson_max	0.9594
spearman_max	0.8606

Training Details

Training Dataset

Unnamed Dataset

Size: 3,192,024 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: 6 tokens
mean: 9.15 tokens
max: 17 tokens

min: 53 tokens
mean: 93.6 tokens
max: 150 tokens

min: 0.0
mean: 0.5
max: 1.0

	sentence_0	sentence_1	label
type	string	string	float
details	min: 6 tokens mean: 9.15 tokens max: 17 tokens	min: 53 tokens mean: 93.6 tokens max: 150 tokens	min: 0.0 mean: 0.5 max: 1.0

Samples:

sentence_0	sentence_1	label
`Must have experience in software development`	Executive Assistant with a strong background in real estate and financial services. Experienced in managing executive schedules, coordinating communications, and supporting investment banking operations. Proficient in office management software and adept at multitasking in fast-paced environments. Previous roles at Blackstone, Piper Sandler, and Broe Real Estate Group, where responsibilities included supporting high-level executives and enhancing operational efficiency. Skilled in fostering relationships and facilitating smooth transitions in fast-paced settings.	`0.0`
`Must have experience in overseeing service delivery for health initiatives`	Director of Solution Strategy in health, wellness, and fitness, specializing in relationship building and strategy execution. Experienced in overseeing service delivery and performance management for telehealth and digital health initiatives at Blue Cross Blue Shield of Massachusetts. Proven track record in vendor lifecycle management, contract strategy, and operational leadership. Skilled in developing standardized wellness programs and enhancing client satisfaction through innovative solutions. Strong background in managing cross-functional teams and driving performance metrics in health engagement and wellness services.	`1.0`
`Must have experience collaborating with Fortune 500 companies`	Senior Sales and Business Development Manager in the energy sector, specializing in increasing profitable sales for small to large companies. Proven track record in relationship building, team management, and strategy development. Experienced in collaborating with diverse stakeholders, including Fortune 500 companies and small to large privately held companies. Previous roles include Vice President of Operations at NovaStar LP and Director of Sales at NovaStar Mortgage and Athlon Solutions. Strong communicator and team player, with a focus on customer needs and operational efficiency.	`1.0`

Loss: CosineSimilarityLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 128
per_device_eval_batch_size: 128
num_train_epochs: 1.0
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 128
per_device_eval_batch_size: 128
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: 1.0
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: 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
eval_use_gather_object: False
batch_sampler: batch_sampler
multi_dataset_batch_sampler: round_robin

Training Logs

Epoch	Step	Training Loss	validation_spearman_max
0.0200	500	0.1362	-
0.0401	1000	0.0533	-
0.0601	1500	0.0433	-
0.0802	2000	0.0386	-
0.1002	2500	0.0356	-
0.1203	3000	0.0345	-
0.1403	3500	0.0326	-
0.1604	4000	0.0323	-
0.1804	4500	0.0313	-
0.2005	5000	0.0305	-
0.2205	5500	0.0298	-
0.2406	6000	0.0296	-
0.2606	6500	0.0291	-
0.2807	7000	0.0286	-
0.3007	7500	0.0286	-
0.3208	8000	0.0281	-
0.3408	8500	0.0278	-
0.3609	9000	0.0273	-
0.3809	9500	0.0276	-
0.4010	10000	0.0274	-
0.4210	10500	0.0266	-
0.4411	11000	0.0261	-
0.4611	11500	0.0264	-
0.4812	12000	0.0256	-
0.5012	12500	0.0254	-
0.5213	13000	0.0251	-
0.5413	13500	0.0249	-
0.5614	14000	0.0253	-
0.5814	14500	0.0247	-
0.6015	15000	0.0254	-
0.6215	15500	0.0246	-
0.6416	16000	0.0251	-
0.6616	16500	0.0248	-
0.6817	17000	0.0247	-
0.7017	17500	0.0246	-
0.7218	18000	0.0242	-
0.7418	18500	0.024	-
0.7619	19000	0.0247	-
0.7819	19500	0.0238	-
0.8020	20000	0.0244	0.8603
0.8220	20500	0.024	-
0.8421	21000	0.0244	-
0.8621	21500	0.0242	-
0.8822	22000	0.0239	-
0.9022	22500	0.0237	-
0.9223	23000	0.0241	-
0.9423	23500	0.0242	-
0.9624	24000	0.0238	-
0.9824	24500	0.0236	-
1.0	24938	-	0.8606

Framework Versions

Python: 3.11.6
Sentence Transformers: 3.0.1
Transformers: 4.44.1
PyTorch: 2.4.0+cu121
Accelerate: 0.33.0
Datasets: 2.21.0
Tokenizers: 0.19.1

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",
}

recruitco
/

embedding_criteria_profile_summary_matching_qa_minilm_v1