blackmlengineer/occucoder-en-venti-v0

SentenceTransformer

This is a sentence-transformers model trained on the train dataset. 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
• Maximum Sequence Length: 32 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Supported Modality: Text
• Training Dataset:
  • train

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': 768, 'pooling_mode': 'mean', 'include_prompt': True})
)

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("blackmlengineer/occucoder-en-venti-v0")
# Run inference
sentences = [
    'Meat Wrapper',
    'Packers and Packagers, Hand',
    'Laundry and Dry-Cleaning Workers',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.6317, 0.1758],
#         [0.6317, 1.0000, 0.1272],
#         [0.1758, 0.1272, 1.0000]])

Training Details

Training Dataset

train

• Dataset: train
• Size: 991,580 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 100 samples:

  	anchor	positive	negative
  type	string	string	string
  modality	text	text	text
  details	min: 3 tokens mean: 4.61 tokens max: 8 tokens	min: 3 tokens mean: 7.7 tokens max: 19 tokens	min: 3 tokens mean: 7.95 tokens max: 22 tokens

• Samples:

  anchor	positive	negative
  Personnel Psychologist	Industrial-Organizational Psychologists	Educational, Guidance, and Career Counselors and Advisors
  Property Claims Adjuster	Claims Adjusters, Examiners, and Investigators	Administrative Law Judges, Adjudicators, and Hearing Officers
  Assistant Professor	Art, Drama, and Music Teachers, Postsecondary	Area, Ethnic, and Cultural Studies Teachers, Postsecondary

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "gather_across_devices": false,
      "directions": [
          "query_to_doc"
      ],
      "partition_mode": "joint",
      "hardness_mode": null,
      "hardness_strength": 0.0
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 128
• num_train_epochs: 5

All Hyperparameters

Click to expand

• per_device_train_batch_size: 128
• num_train_epochs: 5
• 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: 1.0
• label_smoothing_factor: 0.0
• bf16: False
• fp16: False
• 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: 8
• 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
0.0645	500	2.5686
0.1291	1000	1.9404
0.1936	1500	1.6797
0.2582	2000	1.5128
0.3227	2500	1.4104
0.3872	3000	1.3169
0.4518	3500	1.2508
0.5163	4000	1.1896
0.5809	4500	1.1410
0.6454	5000	1.1091
0.7100	5500	1.0743
0.7745	6000	1.0442
0.8390	6500	1.0229
0.9036	7000	0.9909
0.9681	7500	0.9812
1.0327	8000	0.9573
1.0972	8500	0.9400
1.1617	9000	0.9321
1.2263	9500	0.9085
1.2908	10000	0.8988
1.3554	10500	0.8988
1.4199	11000	0.8897
1.4844	11500	0.8875
1.5490	12000	0.8695
1.6135	12500	0.8551
1.6781	13000	0.8574
1.7426	13500	0.8489
1.8072	14000	0.8399
1.8717	14500	0.8389
1.9362	15000	0.8304
2.0008	15500	0.8286
2.0653	16000	0.8100
2.1299	16500	0.8146
2.1944	17000	0.8120
2.2589	17500	0.8063
2.3235	18000	0.8044
2.3880	18500	0.8059
2.4526	19000	0.7951
2.5171	19500	0.7924
2.5816	20000	0.7857
2.6462	20500	0.7888
2.7107	21000	0.7850
2.7753	21500	0.7801
2.8398	22000	0.7813
2.9044	22500	0.7772
2.9689	23000	0.7742
3.0334	23500	0.7706
3.0980	24000	0.7693
3.1625	24500	0.7641
3.2271	25000	0.7543
3.2916	25500	0.7555
3.3561	26000	0.7615
3.4207	26500	0.7583
3.4852	27000	0.7625
3.5498	27500	0.7599
3.6143	28000	0.7529
3.6788	28500	0.7554
3.7434	29000	0.7561
3.8079	29500	0.7523
3.8725	30000	0.7518
3.9370	30500	0.7481
4.0015	31000	0.7459
4.0661	31500	0.7404
4.1306	32000	0.7297
4.1952	32500	0.7408
4.2597	33000	0.7376
4.3243	33500	0.7394
4.3888	34000	0.7422
4.4533	34500	0.7311
4.5179	35000	0.7283
4.5824	35500	0.7356
4.6470	36000	0.7309
4.7115	36500	0.7377
4.7760	37000	0.7353
4.8406	37500	0.7286
4.9051	38000	0.7360
4.9697	38500	0.7317

Training Time

• Training: 4.1 hours

Framework Versions

• Python: 3.14.5
• Sentence Transformers: 5.6.0
• Transformers: 5.12.1
• PyTorch: 2.12.0
• 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",
}

MultipleNegativesRankingLoss

@misc{oord2019representationlearningcontrastivepredictive,
      title={Representation Learning with Contrastive Predictive Coding},
      author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
      year={2019},
      eprint={1807.03748},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/1807.03748},
}