sjmeis/skills-embedding-base-v1

SentenceTransformer

This is a sentence-transformers model trained. 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: 128 tokens
• Output Dimensionality: 768 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': 'ModernBertModel'})
  (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("sentence_transformers_model_id")
# Run inference
sentences = [
    'client service specialist.',
    'Great work environment and the company overall really care about their employees. Good work life balance.',
    'service desk agent.',
]
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, 0.2197, 0.4225],
#         [0.2197, 1.0000, 0.1708],
#         [0.4225, 0.1708, 1.0000]])

Training Details

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 128
• warmup_steps: 0.1
• weight_decay: 0.01
• bf16: True
• per_device_eval_batch_size: 256
• load_best_model_at_end: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• per_device_train_batch_size: 128
• num_train_epochs: 3
• max_steps: -1
• learning_rate: 5e-05
• lr_scheduler_type: linear
• lr_scheduler_kwargs: None
• warmup_steps: 0.1
• optim: adamw_torch_fused
• optim_args: None
• weight_decay: 0.01
• 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: True
• 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: 256
• 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: True
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}
• warmup_ratio: None

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss
0.0209	500	3.6895	-
0.0419	1000	2.3065	-
0.0628	1500	1.8563	-
0.0837	2000	1.6056	-
0.1047	2500	1.4109	-
0.1256	3000	1.2946	-
0.1465	3500	1.2113	-
0.1674	4000	1.1580	-
0.1884	4500	1.0953	-
0.2093	5000	1.0467	-
0.2302	5500	1.0212	-
0.2512	6000	0.9830	-
0.2721	6500	0.9593	-
0.2930	7000	0.9313	-
0.3140	7500	0.9142	-
0.3349	8000	0.8902	-
0.3558	8500	0.8603	-
0.3767	9000	0.8384	-
0.3977	9500	0.8204	-
0.4186	10000	0.8079	1.0798
0.4395	10500	0.7997	-
0.4605	11000	0.7773	-
0.4814	11500	0.7731	-
0.5023	12000	0.7582	-
0.5233	12500	0.7416	-
0.5442	13000	0.7435	-
0.5651	13500	0.7376	-
0.5860	14000	0.7251	-
0.6070	14500	0.7280	-
0.6279	15000	0.7155	-
0.6488	15500	0.7068	-
0.6698	16000	0.7080	-
0.6907	16500	0.7056	-
0.7116	17000	0.6951	-
0.7326	17500	0.6890	-
0.7535	18000	0.6936	-
0.7744	18500	0.6822	-
0.7953	19000	0.6753	-
0.8163	19500	0.6756	-
0.8372	20000	0.6706	0.9167
0.8581	20500	0.6631	-
0.8791	21000	0.6582	-
0.9000	21500	0.6601	-
0.9209	22000	0.6536	-
0.9419	22500	0.6626	-
0.9628	23000	0.6451	-
0.9837	23500	0.6555	-
1.0046	24000	0.6280	-
1.0256	24500	0.6008	-
1.0465	25000	0.5898	-
1.0674	25500	0.5943	-
1.0884	26000	0.5936	-
1.1093	26500	0.5898	-
1.1302	27000	0.5887	-
1.1512	27500	0.5859	-
1.1721	28000	0.5944	-
1.1930	28500	0.5864	-
1.2139	29000	0.5835	-
1.2349	29500	0.5867	-
1.2558	30000	0.5927	0.8757
1.2767	30500	0.5942	-
1.2977	31000	0.5787	-
1.3186	31500	0.5840	-
1.3395	32000	0.5743	-
1.3605	32500	0.5757	-
1.3814	33000	0.5678	-
1.4023	33500	0.5650	-
1.4232	34000	0.5676	-
1.4442	34500	0.5642	-
1.4651	35000	0.5533	-
1.4860	35500	0.5589	-
1.5070	36000	0.5547	-
1.5279	36500	0.5506	-
1.5488	37000	0.5542	-
1.5698	37500	0.5491	-
1.5907	38000	0.5472	-
1.6116	38500	0.5507	-
1.6326	39000	0.5444	-
1.6535	39500	0.5452	-
1.6744	40000	0.5434	0.7943
1.6953	40500	0.5374	-
1.7163	41000	0.5392	-
1.7372	41500	0.5329	-
1.7581	42000	0.5415	-
1.7791	42500	0.5339	-
1.8000	43000	0.5248	-
1.8209	43500	0.5391	-
1.8419	44000	0.5316	-
1.8628	44500	0.5304	-
1.8837	45000	0.5314	-
1.9046	45500	0.5293	-
1.9256	46000	0.5274	-
1.9465	46500	0.5293	-
1.9674	47000	0.5216	-
1.9884	47500	0.5146	-
2.0093	48000	0.5007	-
2.0302	48500	0.4781	-
2.0512	49000	0.4818	-
2.0721	49500	0.4735	-
2.0930	50000	0.4658	0.7542
2.1139	50500	0.4645	-
2.1349	51000	0.4724	-
2.1558	51500	0.4655	-
2.1767	52000	0.4682	-
2.1977	52500	0.4721	-
2.2186	53000	0.4600	-
2.2395	53500	0.4614	-
2.2605	54000	0.4708	-
2.2814	54500	0.4596	-
2.3023	55000	0.4542	-
2.3232	55500	0.4581	-
2.3442	56000	0.4616	-
2.3651	56500	0.4567	-
2.3860	57000	0.4572	-
2.4070	57500	0.4618	-
2.4279	58000	0.4557	-
2.4488	58500	0.4528	-
2.4698	59000	0.4459	-
2.4907	59500	0.4531	-
2.5116	60000	0.4445	0.7255
2.5325	60500	0.4523	-
2.5535	61000	0.4514	-
2.5744	61500	0.4512	-
2.5953	62000	0.4576	-
2.6163	62500	0.4500	-
2.6372	63000	0.4509	-
2.6581	63500	0.4408	-
2.6791	64000	0.4432	-
2.7000	64500	0.4413	-
2.7209	65000	0.4410	-
2.7418	65500	0.4352	-
2.7628	66000	0.4379	-
2.7837	66500	0.4459	-
2.8046	67000	0.4419	-
2.8256	67500	0.4411	-
2.8465	68000	0.4433	-
2.8674	68500	0.4360	-
2.8884	69000	0.4391	-
2.9093	69500	0.4399	-
2.9302	70000	0.4306	0.7062
2.9511	70500	0.4370	-
2.9721	71000	0.4339	-
2.9930	71500	0.4361	-
3.0	71667	-	0.7051

• The bold row denotes the saved checkpoint.

Training Time

• Training: 17.4 hours

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 5.6.0
• Transformers: 5.16.1
• PyTorch: 2.13.0+cu130
• Accelerate: 1.14.0
• Datasets: 5.0.0
• Tokenizers: 0.23.2

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},
}