bge-nuclear-finetuned-4-3 / README.md

robwhelan

Upload fine-tuned BGE embeddings model for nuclear licensing search

fd90831 verified 6 months ago

preview code

raw

history blame contribute delete

18.1 kB

metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:8622
  - loss:MultipleNegativesRankingLoss
base_model: BAAI/bge-base-en-v1.5
widget:
  - source_sentence: >-
      What is the aging management process for concrete structures in nuclear
      power plants?
    sentences:
      - >-
        Loss of material (spalling, scaling) and cracking due to freeze-thaw is
        a significant consideration in the aging management of concrete used for
        structural support and water retaining boundaries in nuclear power
        plants.
      - >-
        The NRC requires thorough inspections of various reactor components to
        ensure compliance with regulatory safety standards.
      - >-
        The applicability of fuel fragmentation and relocation models
        considering the full set of experimental test data presently available
        and higher fuel burnups being proposed in WCAP-18850-P.
  - source_sentence: control rod sequence during reactor startup
    sentences:
      - >-
        With the RWM inoperable during a reactor startup, the operator is still
        capable of enforcing the prescribed control rod sequence. However, the
        overall reliability is reduced because a single operator error can
        result in violating the control rod sequence.
      - >-
        With the RWM inoperable during a reactor shutdown, the operator is still
        capable of enforcing the prescribed control rod sequence.
      - >-
        Potential ecological impacts during aquifer restoration will be SMALL,
        since surface disturbing activities will be limited. Potential impacts
        associated with aquifer restoration activities will include vegetation
        and habitat alteration due to the response and cleanup of potential
        spills, noxious weeds, wildlife mortality from vehicle collisions and
        displacement due to noise, dust, and human/mechanical presence.
        Potential impacts are expected to be less than during operation due to a
        smaller workforce and decreased traffic.
  - source_sentence: What measurement units are used for reactor coolant temperatures?
    sentences:
      - >-
        The coolant temperatures in the reactors are measured in degrees Celsius
        (°C).
      - >-
        The reports frequently summarize operational data over time spans such
        as weeks and months.
      - >-
        MDA: Minimum Detectable Activity refers to the lowest quantity of
        radioactive material that can be reliably detected by a particular
        measurement method.
  - source_sentence: >-
      The NRC mandates specific designs for the safety baskets in boiling water
      reactors to ensure optimal performance.
    sentences:
      - >-
        COBRA-FLX is a computer code developed for the prediction of critical
        heat flux (CHF) in light water reactor thermal-hydraulic systems, and it
        has been validated against a wide range of experimental data to ensure
        its accuracy.
      - >-
        Figure B2-5 presents the schematic of the BWR 81-Assembly Basket, which
        is essential for the containment and organization of nuclear fuel
        assemblies.
      - >-
        The arrangement of different materials in a reactor core affects the
        thermal efficiency greatly, especially under high-pressure conditions.
  - source_sentence: boric acid solution in nuclear reactors
    sentences:
      - >-
        A temperature sensor provides temperature measurement of each tank's
        contents. Temperature indication is provided as well as high and low
        temperature alarms which are indicated on the main control board.
      - >-
        The calculated dose values at different locations are essential for
        ensuring compliance with safety standards.
      - >-
        The concentration of boric acid solution in storage is maintained
        between 4 and 4.4 percent by weight. Periodic manual sampling and
        corrective action, if necessary, assures that tanks are maintained.
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy
model-index:
  - name: SentenceTransformer based on BAAI/bge-base-en-v1.5
    results:
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: validation
          type: validation
        metrics:
          - type: cosine_accuracy
            value: 0.9373840689659119
            name: Cosine Accuracy

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

This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5. 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: BAAI/bge-base-en-v1.5
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 dimensions
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': 512, 'do_lower_case': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, '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 = [
    'boric acid solution in nuclear reactors',
    'The concentration of boric acid solution in storage is maintained between 4 and 4.4 percent by weight. Periodic manual sampling and corrective action, if necessary, assures that tanks are maintained.',
    "A temperature sensor provides temperature measurement of each tank's contents. Temperature indication is provided as well as high and low temperature alarms which are indicated on the main control board.",
]
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: validation
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.9374

Training Details

Training Dataset

Unnamed Dataset

Size: 8,622 training samples
Columns: sentence_0, sentence_1, and sentence_2

Approximate statistics based on the first 1000 samples:

	sentence_0	sentence_1	sentence_2
type	string	string	string
details	min: 4 tokens mean: 14.85 tokens max: 41 tokens	min: 6 tokens mean: 43.91 tokens max: 336 tokens	min: 4 tokens mean: 30.45 tokens max: 228 tokens

Samples:

sentence_0	sentence_1	sentence_2
`The NRC mandates monitoring and calibration of voltage outputs during reactor startup procedures.`	`Simulates pressing GEN FIELD FLASH pushbutton until Generator voltage stops rising. (GENERATOR LINE VOLTS should indicate ~260V).`	`Operator reads the Note and continues.`
`What are the environmental impacts of nuclear facilities during decommissioning?`	`The Environmental Impacts Comparison Summary provides a detailed analysis of various impact areas associated with the proposed action compared to the termination of operations and decommissioning of nuclear facilities.`	`The report outlines the thermal efficiency of natural gas combined cycle (NGCC) power plants, which can influence operational costs and the decision-making process for future energy projects.`
`ECCS signal path`	`The ECCS/NSSSS Signal Path is crucial for the timely activation of emergency core cooling systems to prevent overheating of the reactor core during an accident.`	`The integration of signal pathways within reactor control systems ensures effective monitoring and response capability, which is essential for maintaining safety standards in nuclear operations.`

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: 32
per_device_eval_batch_size: 32
num_train_epochs: 5
fp16: True
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 32
per_device_eval_batch_size: 32
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
num_train_epochs: 5
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
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}
tp_size: 0
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
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: batch_sampler
multi_dataset_batch_sampler: round_robin

Training Logs

Epoch	Step	Training Loss	validation_cosine_accuracy
0.7407	200	-	0.9276
1.0	270	-	0.9309
1.4815	400	-	0.9314
1.8519	500	0.5662	-
2.0	540	-	0.9355
2.2222	600	-	0.9355
2.9630	800	-	0.9374

Framework Versions

Python: 3.10.14
Sentence Transformers: 4.1.0
Transformers: 4.51.3
PyTorch: 2.2.2
Accelerate: 1.6.0
Datasets: 3.5.0
Tokenizers: 0.21.1

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}
}