brilan/procedure-tool-matching_10_epochs

SentenceTransformer based on sentence-transformers/all-mpnet-base-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-mpnet-base-v2. It maps sentences & paragraphs to a 768-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/all-mpnet-base-v2
• Maximum Sequence Length: 384 tokens
• Output Dimensionality: 768 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': 384, 'do_lower_case': False}) with Transformer model: MPNetModel 
  (1): Pooling({'word_embedding_dimension': 768, '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("brilan/procedure-tool-matching_10_epochs")
# Run inference
sentences = [
    'list running processes',
    'Displays information about services and drivers on a local or remote computer.',
    'Displays the directory structure of a path or of the disk in a drive graphically.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Triplet

• Dataset: dev
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.8614
dot_accuracy	0.1386
manhattan_accuracy	0.856
euclidean_accuracy	0.8614
max_accuracy	0.8614

Triplet

• Dataset: dev
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	1.0
dot_accuracy	0.0
manhattan_accuracy	1.0
euclidean_accuracy	1.0
max_accuracy	1.0

Triplet

• Dataset: test
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	1.0
dot_accuracy	0.0
manhattan_accuracy	1.0
euclidean_accuracy	1.0
max_accuracy	1.0

Training Details

Training Dataset

Unnamed Dataset

• Size: 7,385 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 5 tokens mean: 9.6 tokens max: 17 tokens	min: 5 tokens mean: 18.53 tokens max: 47 tokens	min: 5 tokens mean: 17.9 tokens max: 57 tokens

• Samples:

  anchor	positive	negative
  added user accounts to the User and Admin groups	use to create a new local user account on a Windows system.	Adds a new subkey or entry to the registry.
  get cached credentials	manipulate privilege on process.	Use to display a list of computers and shared resources on a network.
  used compromised domain accounts to gain access to the target environment	allows users to execute commands remotely on target systems using various methods including WMI, SMB, SSH, RDP, and PowerShell	Copies files and directories including subdirectories.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Evaluation Dataset

Unnamed Dataset

• Size: 1,847 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 5 tokens mean: 9.74 tokens max: 17 tokens	min: 5 tokens mean: 18.21 tokens max: 47 tokens	min: 5 tokens mean: 18.3 tokens max: 57 tokens

• Samples:

  anchor	positive	negative
  obtain information about the domain	It retrieves a list of current network connections.	Saves a copy of specified subkeys, entries, and values of the registry in a specified file.
  obtain credentials from Vault files	retrieve stored passwords from various software and operating systems	allows users to execute commands remotely on target systems using various methods including WMI, SMB, SSH, RDP, and PowerShell
  obtain information about the domain	Get user name and group information along with the respective security identifiers (SID) claims privileges logon identifier (logon ID) for the current user on the local system.	Creates a new shadow copy of a specified volume.

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 10
• warmup_ratio: 0.1
• fp16: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• 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: 1.0
• num_train_epochs: 10
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• 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: True
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	dev_cosine_accuracy	dev_max_accuracy	test_cosine_accuracy
0	0	-	-	0.8614	0.8614	-
0.2165	100	2.3461	1.3114	0.9995	-	-
0.4329	200	1.5881	1.2268	0.9995	-	-
0.6494	300	1.5293	1.2106	0.9995	-	-
0.8658	400	1.4955	1.1909	1.0	-	-
1.0823	500	0.8778	1.2624	0.9978	-	-
1.2987	600	0.0	1.2644	0.9989	-	-
1.2143	700	1.4556	1.1650	1.0	-	-
1.4307	800	1.4347	1.1479	1.0	-	-
1.6472	900	1.4549	1.1127	1.0	-	-
1.8636	1000	1.4315	1.1446	1.0	-	-
2.0801	1100	0.8624	1.1487	1.0	-	-
2.2965	1200	0.0004	1.1511	0.9984	-	-
2.2121	1300	1.3961	1.1081	1.0	-	-
2.4286	1400	1.4033	1.1076	1.0	-	-
2.6450	1500	1.4211	1.1022	1.0	-	-
2.8615	1600	1.4044	1.1364	1.0	-	-
3.0779	1700	0.864	1.1135	1.0	-	-
3.2944	1800	0.0005	1.1156	1.0	-	-
3.2100	1900	1.353	1.0914	1.0	-	-
3.4264	2000	1.3805	1.0958	1.0	-	-
3.6429	2100	1.4068	1.0925	1.0	-	-
3.8593	2200	1.3874	1.1184	1.0	-	-
4.0758	2300	0.8734	1.0992	1.0	-	-
4.2922	2400	0.0011	1.1007	1.0	-	-
4.2078	2500	1.3287	1.0853	1.0	-	-
4.4242	2600	1.3691	1.0944	1.0	-	-
4.6407	2700	1.4026	1.0906	1.0	-	-
4.8571	2800	1.3816	1.0926	1.0	-	-
5.0736	2900	0.8775	1.0915	1.0	-	-
5.2900	3000	0.0007	1.0924	1.0	-	-
5.2056	3100	1.3095	1.0838	1.0	-	-
5.4221	3200	1.3669	1.0875	1.0	-	-
5.6385	3300	1.389	1.0869	1.0	-	-
5.8550	3400	1.3741	1.0835	1.0	-	-
6.0714	3500	0.8852	1.0864	1.0	-	-
6.2879	3600	0.0005	1.0866	1.0	-	-
6.2035	3700	1.2937	1.0793	1.0	-	-
6.4199	3800	1.3618	1.0852	1.0	-	-
6.6364	3900	1.3848	1.0847	1.0	-	-
6.8528	4000	1.3722	1.0799	1.0	-	-
7.0693	4100	0.8995	1.0827	1.0	-	-
7.2857	4200	0.0006	1.0826	1.0	-	-
7.2013	4300	1.2766	1.0775	1.0	-	-
7.4177	4400	1.3559	1.0791	1.0	-	-
7.6342	4500	1.3806	1.0793	1.0	-	-
7.8506	4600	1.3636	1.0788	1.0	-	-
7.8939	4620	-	-	-	-	1.0

Framework Versions

• Python: 3.11.5
• Sentence Transformers: 3.1.0
• Transformers: 4.44.2
• PyTorch: 2.4.1+cu121
• Accelerate: 1.0.0
• Datasets: 3.0.1
• 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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}