SentenceTransformer based on BAAI/bge-base-en-v1.5
This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for retrieval.
Model Details
Model Description
- Model Type: Sentence Transformer
- Base model: BAAI/bge-base-en-v1.5
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 768 dimensions
- Similarity Function: Cosine Similarity
- Supported Modality: Text
Model Sources
Full Model Architecture
SentenceTransformer(
(0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'BertModel'})
(1): Pooling({'embedding_dimension': 768, 'pooling_mode': 'cls', '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
model = SentenceTransformer("sentence_transformers_model_id")
sentences = [
'Professional SummaryArdent and intelligent software engineer capable of providing sterling service using my technical prowess, hard working nature and interpersonal skills.\nSkillsJava, C, C++, XML, HTML, CSS, JavaScript, Spring MVC, RESTful Web Services, J2EE-JSP, Servlet, JDBC, SQL - MySQL, Oracle, SQL Server, Hadoop-MapReduce, Hive. · MATLAB-programming and Modeling, Neural Network Regression Tool, Genetic Programming GPTIPS Tool.Proficient in developing Machine Learning Algorithms viz. Neural Network, Genetic Algorithms, Genetic Programming, Swarm intelligence based algorithms. · Microsoft Excel Functions, Regression Tool Box and VBA for Data cleansing, normalizing and scaling . · VLSI Chip Designing using CADENCE Tool. · Windows, Linux, Unix. · Eclipse, MS Office, Latex, Visual Studio\nWork History12/2015toPresentSoftware DeveloperRapid Global Business Solutions, Inc–,,Client: State Teachers Retirement System, Ohio.Technologies Used: Java, Spring and Hibernate Frameworks, GWT Framework, RESTful Web Services, XML, JSON, JDBC, JSP, HTML, CSS, Oracle, PL/SQL.08/2013to08/2014Graduate Teaching AssistantPurdue University–,,for the courses: Objected Oriented Programming and Data Structures, Network Security, Data Driven Websites, Engineering Calculus, Unix and C.Grader for the.\nEducationExpected inAugust, 2015totoM.S:Computer Science and EngineeringUniversity of Toledo-,GPA:Computer Science and Engineering GPA: 3.83 / 4.0Expected in2013totoB. E:Electrical and Electronics EngineeringAnna University-,GPA:Electrical and Electronics Engineering GPA: 8.32 / 10.0\nSkillsC++, Calculus, Eclipse, XML, Java, JSP, JavaScript, JDBC, JSON, Linux, Machine Learning, Math, MATLAB, MySQL, Network Security, Oracle, PL/SQL, scaling, SQL, SQL Server, Unix and C, Unix, VBA, Visual Studio',
"Who we are:Founded in 2017, Gatik is the leader in autonomous middle mile logistics. We deliver goods safely and efficiently using medium duty trucks with a focus on short-haul, B2B logistics for Fortune 500 customers such as Walmart and Loblaw. Gatik enables our customers to optimize their hub-and-spoke operations, enhance service levels and product flow across multiple locations while reducing labor costs and meeting an unprecedented expectation for faster deliveries.\nAbout the role:We are seeking passionate Staff Software Engineers, who have strong fundamentals in software development practices and are experts in C++ language in production-oriented environment. The ideal candidate is a highly experienced C++ developer with a passion for enabling the world's first safe, reliable & efficient network of autonomous vehicles. You will partner with the research and software engineers to design, develop, test and validate AV features for our autonomous fleet.\nThis role will be onsite 4 days a week at our Mountain View office!\nWhat you'll do:Design, implement, integrate, and support real-time mission-critical software for the Gatiks autonomy stackWork with the research engineers to develop maintainable, testable and robust software designsArchitect and implement solutions to complex issues between components partitioned across the large software stackBe at the forefront of guiding & ensuring best SDLC practices while contributing to improving the safety in the core autonomy stackCollaborate with the Infrastructure and DevOps teams for efficient, secure and scalable software delivery to a network of Gatiks autonomous fleetGuide and mentor autonomy researchers and algorithm developers to make sure their components are running efficiently and with optimal compute and memory usageReview and refine technical requirements and translate them into high-level design & plans to support the development of safe AV technologyConduct code and design reviews and advise on technical mattersWhat we're looking for:Bachelor's Degree in Computer Science, Robotics or in a related degreeMaster's or PhD degree preferred7+ years relevant industry work experience in a production environmentExpert software developer in C++ and build systems (Conan, Cmake, Bazel)Experience using software project management software (Jira, Confluence, etc.)Proven system integration and software architecture skillsStrong background in linernon-liner optimization, linear algebra or statisticsExperience with building frameworkssoftware infrastructure for large scale projects\nMore about Gatik:With headquarters in Mountain View, CA and offices in Canada, Texas, Louisiana and Arkansas, Gatik is establishing new standards of success for the autonomous trucking industry every day. Visit us at Gatik for more company information and Jobs @ Gatik for more open roles.\nGatik News:The 10 most innovative companies in transportation of 2023Americas Best Startup Employers of 2023 by ForbesAmericas Best Startup Employers of 2022 by ForbesGatik been named as a 2023 FreightTech 25 winner!Gatik named a TIME Best Invention of 2022Apeksha Kumavat recognized on the Inc. 2022 Female Founders 100 ListGatiks Gautam Narang on the importance of knowing your customerGatik CEO Gautam Narang was Named Among the Most Exceptional Entrepreneurs of 2022 by Goldman Sachs\nTaking care of our team:At Gatik, we connect people of extraordinary talent and experience to an opportunity to create a more resilient supply chain and contribute to our environments sustainability. We are diverse in our backgrounds and perspectives yet united by a bold vision and shared commitment to our values. Our culture emphasizes the importance of collaboration, respect and agility. We at Gatik strive to create a diverse and inclusive environment where everyone feels they have opportunities to succeed and grow because we know that together we can do great things. We are committed to an inclusive and diverse team. We do not discriminate based on race, color, ethnicity, ancestry, national origin, religion, sex, gender, gender identity, gender expression, sexual orientation, age, disability, veteran status, genetic information, marital status or any legally protected status.",
"Novi Labs delivers world-class Oil & Gas forecasts via an end-to-end platform that leverages machine learning and data science. Join our talented technical team to help us improve and expand our products and further solidify our market-leading position. \n\nBenefits of Working at Novi Labs\nSmall, talented teamHighly collaborative work environment High autonomy with true project ownershipSmall company with visibility into company decisions and prioritiesEasy line-of-sight to how your contributions directly benefit customers\n\nRole\nWe are looking for a seasoned data engineer that has a passion for sound design and engineering practices and is adept at making reasoned short-term vs long-term trade-offs. In this role you will be leading the buildout of our next generation data and machine learning platform, while helping to support and transition from our current systems. This is a key role on a small team. To be successful, It will be important that you feel comfortable driving projects on your own, but also enjoy brainstorming and collaboration. \nWhat youll do here\nAfter joining Novi, you'll collaborate hand in hand with our team of data scientists and ETL specialists. Together, we're charting a course towards establishing an innovative greenfield platform that will unlock the power of machine-learning powered analytics products.\nAs a key member of our team, you'll play a pivotal role in cultivating a culture of engineering excellence. Your responsibilities will encompass the establishment of standards and the integration of emerging technologies to uphold the zenith of software engineering practices in Python. Your expertise will contribute to the improvement of our existing analytics platform, encompassing the implementation of novel functionalities and the enhancement of both the scalability and reliability of our data transformation and storage. This will be achieved through precise profiling and optimization techniques, coupled with the strategic design of cloud computing solutions.\nIf you're driven by the prospect of making a profound impact within a collaborative and forward-thinking environment, we invite you to join us at Novi. Together, we'll continue to elevate the benchmark of data-driven solutions while nurturing your professional growth and development.\nRequired Skills & Experience\nMinimum of 5 years of experience building robust and scalable data pipelinesStrong foundation in software engineering principles in PythonKnowledge of the PyData stack (e.g. pandas) that goes deeper than the APIsExperience handling, storing and linking multi-modal data typesExperience with the full software development lifecycle (git, CI, Docker, test, etc.)Strong database knowledge (SQL & NoSQL)Strong cloud architecture skills - AWS preferred\n\nDesired Skills & Experience\nInterest in data science or previous experience working on a data science teamExperience working with GIS dataPrevious work on an agile teamExperience working for a start-up or small company\n\n\n\nCultural Fit\nAt Novi, the culture fit is as important as the technical fit. Our team will be a great fit for you if you:\nHave a positive attitude and a growth mindsetEnjoy working on small, highly collaborative teamsAre self-motivated to improve your technical capacitiesInfluence without authority via well-reasoned ideas and clear communicationAre comfortable working cross-functionally to deliver customer impactEnjoy team and company successes as much as your ownTake ownership of deliverables with strong attention to detailAre happy to task switch when necessary to help teams succeedEnjoy knowing whats happening across the entire company\n\nMore About Novi\nProfessional work environment with a clear understanding of worklife balanceFlexible Leave PolicyHeathDentalVision Insurance401K\n",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
similarities = model.similarity(embeddings, embeddings)
print(similarities)
Evaluation
Metrics
Semantic Similarity
| Metric |
Value |
| pearson_cosine |
0.5518 |
| spearman_cosine |
0.5657 |
Binary Classification
| Metric |
Value |
| cosine_accuracy |
0.7782 |
| cosine_accuracy_threshold |
0.8133 |
| cosine_f1 |
0.7937 |
| cosine_f1_threshold |
0.8065 |
| cosine_precision |
0.7555 |
| cosine_recall |
0.8359 |
| cosine_ap |
0.8222 |
| cosine_mcc |
0.5558 |
Training Details
Training Dataset
Unnamed Dataset
Evaluation Dataset
Unnamed Dataset
Training Hyperparameters
Non-Default Hyperparameters
per_device_train_batch_size: 16
num_train_epochs: 6
learning_rate: 1e-05
warmup_steps: 0.1
weight_decay: 0.05
fp16: True
per_device_eval_batch_size: 16
All Hyperparameters
Click to expand
per_device_train_batch_size: 16
num_train_epochs: 6
max_steps: -1
learning_rate: 1e-05
lr_scheduler_type: linear
lr_scheduler_kwargs: None
warmup_steps: 0.1
optim: adamw_torch_fused
optim_args: None
weight_decay: 0.05
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
optim_target_modules: None
gradient_accumulation_steps: 1
average_tokens_across_devices: True
max_grad_norm: 1.0
label_smoothing_factor: 0.0
bf16: False
fp16: True
bf16_full_eval: False
fp16_full_eval: False
tf32: None
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
use_liger_kernel: False
liger_kernel_config: None
use_cache: False
neftune_noise_alpha: None
torch_empty_cache_steps: None
auto_find_batch_size: False
log_on_each_node: True
logging_nan_inf_filter: True
include_num_input_tokens_seen: no
log_level: passive
log_level_replica: warning
disable_tqdm: False
project: huggingface
trackio_space_id: None
trackio_bucket_id: None
trackio_static_space_id: None
per_device_eval_batch_size: 16
prediction_loss_only: True
eval_on_start: False
eval_do_concat_batches: True
eval_use_gather_object: False
eval_accumulation_steps: None
include_for_metrics: []
batch_eval_metrics: False
save_only_model: False
save_on_each_node: False
enable_jit_checkpoint: False
push_to_hub: False
hub_private_repo: None
hub_model_id: None
hub_strategy: every_save
hub_always_push: False
hub_revision: None
load_best_model_at_end: False
ignore_data_skip: False
restore_callback_states_from_checkpoint: False
full_determinism: False
seed: 42
data_seed: None
use_cpu: False
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
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_pin_memory: True
dataloader_persistent_workers: False
dataloader_prefetch_factor: None
remove_unused_columns: True
label_names: None
train_sampling_strategy: random
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
ddp_static_graph: None
ddp_backend: None
ddp_timeout: 1800
fsdp: None
fsdp_config: None
deepspeed: None
debug: []
skip_memory_metrics: True
do_predict: False
resume_from_checkpoint: None
warmup_ratio: None
local_rank: -1
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}
Training Logs
| Epoch |
Step |
Training Loss |
Validation Loss |
resume-job-sim_spearman_cosine |
resume-job-bin_cosine_ap |
| 0.1279 |
50 |
0.0339 |
0.0293 |
0.2505 |
0.6404 |
| 0.2558 |
100 |
0.0285 |
0.0286 |
0.3005 |
0.6645 |
| 0.3836 |
150 |
0.0260 |
0.0270 |
0.3440 |
0.6802 |
| 0.5115 |
200 |
0.0247 |
0.0259 |
0.3822 |
0.7049 |
| 0.6394 |
250 |
0.0237 |
0.0238 |
0.4618 |
0.7488 |
| 0.7673 |
300 |
0.0229 |
0.0233 |
0.4630 |
0.7489 |
| 0.8951 |
350 |
0.0232 |
0.0227 |
0.4882 |
0.7620 |
| 1.0230 |
400 |
0.0212 |
0.0237 |
0.4907 |
0.7623 |
| 1.1509 |
450 |
0.0189 |
0.0225 |
0.5100 |
0.7685 |
| 1.2788 |
500 |
0.0201 |
0.0219 |
0.5139 |
0.7646 |
| 1.4066 |
550 |
0.0198 |
0.0223 |
0.5186 |
0.7833 |
| 1.5345 |
600 |
0.0185 |
0.0225 |
0.5209 |
0.7820 |
| 1.6624 |
650 |
0.0184 |
0.0220 |
0.5324 |
0.7928 |
| 1.7903 |
700 |
0.0187 |
0.0221 |
0.5291 |
0.7887 |
| 1.9182 |
750 |
0.0182 |
0.0218 |
0.5322 |
0.7913 |
| 2.0460 |
800 |
0.0161 |
0.0222 |
0.5400 |
0.8002 |
| 2.1739 |
850 |
0.0172 |
0.0228 |
0.5460 |
0.8094 |
| 2.3018 |
900 |
0.0157 |
0.0223 |
0.5453 |
0.8050 |
| 2.4297 |
950 |
0.0151 |
0.0227 |
0.5428 |
0.8024 |
| 2.5575 |
1000 |
0.0164 |
0.0221 |
0.5652 |
0.8193 |
| 2.6854 |
1050 |
0.0160 |
0.0221 |
0.5624 |
0.8156 |
| 2.8133 |
1100 |
0.0148 |
0.0218 |
0.5766 |
0.8253 |
| 2.9412 |
1150 |
0.0154 |
0.0224 |
0.5657 |
0.8222 |
Training Time
Framework Versions
- Python: 3.12.13
- Sentence Transformers: 5.6.0
- Transformers: 5.12.1
- PyTorch: 2.11.0+cu128
- Accelerate: 1.14.0
- Datasets: 5.0.0
- Tokenizers: 0.22.2
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",
}
ContrastiveLoss
@inproceedings{hadsell2006dimensionality,
author={Hadsell, R. and Chopra, S. and LeCun, Y.},
booktitle={2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06)},
title={Dimensionality Reduction by Learning an Invariant Mapping},
year={2006},
volume={2},
number={},
pages={1735-1742},
doi={10.1109/CVPR.2006.100}
}