dinho1597/bge-small-qa-telecom-ft

SentenceTransformer based on BAAI/bge-small-en-v1.5

This is a sentence-transformers model finetuned from BAAI/bge-small-en-v1.5 on the telecom-qa-multiple_choice dataset. It maps sentences & paragraphs to a 384-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: BAAI/bge-small-en-v1.5
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • telecom-qa-multiple_choice

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': 512, 'do_lower_case': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'What should the AP or PCP do if it is not decentralized AP or PCP clustering capable or a decentralized AP or PCP cluster is not present?',
    'If the AP or PCP is not decentralized AP or PCP clustering capable or a decentralized AP or PCP cluster is not present, it should set its Cluster Member Role to 0 (not currently participating in a cluster) and remain unclustered.',
    'When a Data, Management or Extension frame is received, a STA inserts it in an appropriate cache.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

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

Evaluation

Metrics

Information Retrieval

• Dataset: telecom-ir-eval
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.9657
cosine_accuracy@3	0.9924
cosine_accuracy@5	0.9931
cosine_accuracy@10	0.9939
cosine_precision@1	0.9657
cosine_recall@1	0.9657
cosine_ndcg@10	0.9824
cosine_mrr@10	0.9784
cosine_map@100	0.9786

Training Details

Training Dataset

telecom-qa-multiple_choice

• Dataset: telecom-qa-multiple_choice at 73aebbb
• Size: 6,552 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 4 tokens mean: 18.95 tokens max: 49 tokens	min: 9 tokens mean: 29.33 tokens max: 112 tokens

• Samples:

  anchor	positive
  What is the goal of a jammer in a mobile edge caching system?	The goal of a jammer in a mobile edge caching system is to interrupt ongoing radio transmissions of the edge node with cached chunks or caching users and prevent access to cached content. Additionally, jammers aim to deplete the resources of edge nodes, caching users, and sensors during failed communication attempts.
  Which type of DRL uses DNNs (Deep Neural Networks) to fit action values and employs experience replay and target networks to ensure stable training convergence?	Value-based DRL, such as Deep Q-Learning (DQL), uses DNNs to fit action values and employs experience replay and target networks to ensure stable training convergence.
  What is the relationship between the curvature of the decision boundary and the robustness of a network?	The lower the curvature of the decision boundaries, the more robust the network.

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

telecom-qa-multiple_choice

• Dataset: telecom-qa-multiple_choice at 73aebbb
• Size: 6,552 evaluation samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 4 tokens mean: 18.87 tokens max: 56 tokens	min: 8 tokens mean: 29.45 tokens max: 91 tokens

• Samples:

  anchor	positive
  Which forward error correction (FEC) codes are available for the THz single carrier mode?	The THz single carrier mode (THz-SC PHY) in the IEEE 802.15.3d standard supports two low-density parity-check (LDPC) codes: 14/15 LDPC (1440,1344) and 11/15 LDPC (1440,1056).
  Which multiple access technique allows users to access the channel simultaneously using the same frequency and time resources, with different power levels?	Non-Orthogonal Multiple Access (NOMA) allows users to access the channel simultaneously using the same frequency and time resources, but with different power levels.
  What is the power gain when doubling the number of antennas?	Doubling the number of antennas yields a 3-dB power gain.

• 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: 512
• per_device_eval_batch_size: 512
• weight_decay: 0.01
• num_train_epochs: 15
• lr_scheduler_type: cosine_with_restarts
• warmup_ratio: 0.1
• fp16: True
• load_best_model_at_end: 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: 512
• per_device_eval_batch_size: 512
• 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: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 15
• max_steps: -1
• lr_scheduler_type: cosine_with_restarts
• 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: True
• 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: None
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• include_for_metrics: []
• 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
• use_liger_kernel: False
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	telecom-ir-eval_cosine_ndcg@10
1.2727	15	1.0332	0.0968	0.9725
2.5455	30	0.2091	0.0518	0.9808
3.8182	45	0.0997	0.0470	0.9824

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.3.1
• Transformers: 4.47.1
• PyTorch: 2.5.1+cu121
• Accelerate: 1.2.1
• Datasets: 3.2.0
• Tokenizers: 0.21.0

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}
}