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

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, 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
• Supported Modality: Text

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

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': 384, 'pooling_mode': 'mean', '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("kowshiktg/recruiter-engine-minilm")
# Run inference
sentences = [
    'ROLE - AI Engineer',
    "Title: Project Manager\nHeadline: Project Manager | Helping teams scale\nSummary: Professional with 3.9+ years of experience. I'm a marketing manager with substantial experience in cross-functional collaboration, stakeholder management, and execution. Lately I've been curious about how AI tools could augment my work — I've experimented with ChatGPT and a few other tools for productivity and content creation, and I think the space is exciting. Open to roles where I can apply my domain expertise alongside emerging AI capabilities.\nLocation: Kolkata, West Bengal\nYears of experience: 3.9\nSkills: PowerPoint, Figma, CI/CD, Illustrator, FastAPI, GCP, BigQuery, Agile\nCareer: Project Manager at Initech. Operations management role at a logistics company. Owned daily fulfillment operations across 3 warehouses, managing a team of 80 across receiving, picking, packing, and outbound. Built and tracked the operational KPIs (on-time fulfillment, accuracy, cost per order) and led the continuous improvement initiatives that drove a 22% productivity gain over 18 months.\nCareer: Project Manager at Pied Piper. Operations management role at a logistics company. Owned daily fulfillment operations across 3 warehouses, managing a team of 80 across receiving, picking, packing, and outbound. Built and tracked the operational KPIs (on-time fulfillment, accuracy, cost per order) and led the continuous improvement initiatives that drove a 22% productivity gain over 18 months.\nCareer: Sales Executive at Stark Industries. Senior accounting role at a mid-sized company — month-end close, financial reporting, statutory compliance (GAAP / Ind-AS), and tax filings. Owned the GL, fixed-asset register, and the audit-readiness function. Managed a team of 3 staff accountants. Built strong process discipline around the close cycle, reducing close time from 12 days to 7 over the last two years.\nEducation: B.Sc Artificial Intelligence at Symbiosis International",
    "Title: Mobile Developer\nHeadline: Mobile Developer | Full-stack development\nSummary: Software engineer with 5.5 years of experience across web, backend, and cloud systems. Strong fundamentals in software development and system design. I've worked across web frontends, REST APIs, and cloud deployments; comfortable in most parts of a typical SaaS stack. I've been keeping up with AI/ML at a self-learner level — taken some online courses, played with the OpenAI and Anthropic APIs, built a small RAG side project — but I haven't done it in a professional capacity yet. Open to roles where I can either deepen my software engineering work or, if the team is open to it, start contributing to ML-adjacent systems.\nLocation: Chennai, Tamil Nadu\nYears of experience: 5.5\nSkills: GraphQL, Salesforce CRM, Tally, Java, GCP, Spark, Time Series, dbt, Computer Vision, Photoshop, React, Illustrator, Hadoop, PostgreSQL, TTS, Fine-tuning LLMs\nCareer: Mobile Developer at Flipkart. Android mobile development using Java and (more recently) Kotlin at a consumer-app company. Built and maintained multiple production features including the main shopping flow, push notification system, and the offline-first sync layer. Comfortable with the Android framework, Jetpack components, and the typical patterns (MVVM, Hilt, Coroutines). My career has been entirely on mobile so far; interested in expanding into broader backend or platform engineering.\nCareer: .NET Developer at Dunder Mifflin. Cloud infrastructure and DevOps work at an enterprise SaaS company. Owned the AWS account architecture (VPC, IAM, networking), the Terraform modules for our service deployments, and the Kubernetes cluster operations. Designed the CI/CD pipelines (GitLab CI + ArgoCD) and the monitoring stack (Prometheus, Grafana, Loki). Strong on the infra and ops side; haven't done much application development.\nCareer: Cloud Engineer at Capgemini. Java backend development at a large enterprise — Spring Boot microservices, Kafka for inter-service messaging, Postgres + Redis for storage. Worked on the customer onboarding flow which involved orchestrating multiple downstream services. Solid on the Spring ecosystem, transaction handling, and the operational side of Java services. Looking to either go deeper on distributed systems or expand into modern application stacks.\nEducation: M.Tech Civil Engineering at VIT Vellore",
]
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, 0.3702, 0.4254],
#         [0.3702, 1.0000, 0.5617],
#         [0.4254, 0.5617, 1.0000]])

Training Details

Training Dataset

Candidate Dataset

• Size: 8,500 training samples
• Columns: sentence_0, sentence_1, and label
• Approximate statistics based on the first 100 samples:

  	sentence_0	sentence_1	label
  type	string	string	float
  modality	text	text
  details	min: 6 tokensmean: 6.0 tokensmax: 6 tokens	min: 238 tokensmean: 255.57 tokensmax: 256 tokens	min: 0.06mean: 0.41max: 0.63

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss",
      "cos_score_transformation": "torch.nn.modules.linear.Identity"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 128
• num_train_epochs: 2
• fp16: True
• per_device_eval_batch_size: 128
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

• per_device_train_batch_size: 128
• num_train_epochs: 2
• max_steps: -1
• learning_rate: 5e-05
• lr_scheduler_type: linear
• lr_scheduler_kwargs: None
• warmup_steps: 0
• optim: adamw_torch_fused
• optim_args: None
• weight_decay: 0.0
• 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
• 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: 128
• 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
• dataloader_multiprocessing_context: None
• dataloader_in_order: True
• 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
• local_rank: -1
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin
• router_mapping: {}
• learning_rate_mapping: {}
• warmup_ratio: None

Training Time

• End-to-end experiment time: 6+ hours
• Fine-tuning time: 2.2 minutes

Framework Versions

• Python: 3.13.15
• Sentence Transformers: 5.7.0
• Transformers: 5.16.1
• PyTorch: 2.11.0+cu128
• Accelerate: 1.14.0
• Datasets: 4.0.0
• Tokenizers: 0.23.1

Additional Resources

• Training and Finetuning Embedding Models with Sentence Transformers: the end-to-end guide for training or finetuning Sentence Transformer models.
• Introduction to Matryoshka Embedding Models: variable-size embeddings that can be truncated with minimal quality loss.
• Binary and Scalar Embedding Quantization for Significantly Faster & Cheaper Retrieval: post-training compression of embedding vectors.
• Multimodal Embedding & Reranker Models with Sentence Transformers: use text, image, audio, and video models through the same API.
• Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers: train multimodal embedding models, with a Visual Document Retrieval walkthrough.

Data Attribution

Training examples were derived from hiring-domain candidate data used in the Redrob competition.

Credit:

Redrob AI / Redrob competition

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",
}