metadata

base_model: google/flan-t5-base
datasets:
  - PiC/phrase_similarity
language:
  - en
library_name: sentence-transformers
metrics:
  - cosine_accuracy
  - cosine_accuracy_threshold
  - cosine_f1
  - cosine_f1_threshold
  - cosine_precision
  - cosine_recall
  - cosine_ap
  - dot_accuracy
  - dot_accuracy_threshold
  - dot_f1
  - dot_f1_threshold
  - dot_precision
  - dot_recall
  - dot_ap
  - manhattan_accuracy
  - manhattan_accuracy_threshold
  - manhattan_f1
  - manhattan_f1_threshold
  - manhattan_precision
  - manhattan_recall
  - manhattan_ap
  - euclidean_accuracy
  - euclidean_accuracy_threshold
  - euclidean_f1
  - euclidean_f1_threshold
  - euclidean_precision
  - euclidean_recall
  - euclidean_ap
  - max_accuracy
  - max_accuracy_threshold
  - max_f1
  - max_f1_threshold
  - max_precision
  - max_recall
  - max_ap
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:7004
  - loss:SoftmaxLoss
widget:
  - source_sentence: >-
      The valve will open 100% when the set point is reached and will remain
      open until a certain blow down factor is reached.
    sentences:
      - >-
        Having raised $17,000,000 in a standard matter, one of the first
        speculative IPOs, Tucker needed more money to continue development of
        the car.
      - >-
        The valve will open 100% when the tennis scoring protocol is reached and
        will remain open until a certain blow down factor is reached.
      - >-
        But the government of PML (N) gave it the complete exponential of a
        Tehsil.
  - source_sentence: >-
      Java BluePrints was the first source to promote Model View Controller
      (MVC) and Data Access Object (DAO) for Java EE application development.
    sentences:
      - >-
        Java BluePrints was the pioneer authority to promote Model View
        Controller (MVC) and Data Access Object (DAO) for Java EE application
        development.
      - >-
        One of the primary job of IIUG is to publish news through a monthly
        newsletter ("The Insider").
      - >-
        Opera Dragonfly must be downloaded on original practice, and functions
        offline thereafter.
  - source_sentence: It also appears immediately after the first shower of the monsoon.
    sentences:
      - >-
        The latter can be minimised by meticulous precision to the wheel
        bearings, tyre sizes and pressures, and brakes (to avoid parasitic brake
        drag).
      - It also appears immediately after the initial rain of the monsoon.
      - >-
        McCullough filed a second appeal that could not be denied without a
        hearing from the State Attorney's Office.
  - source_sentence: >-
      This type places the shifters closer to the hand positions, but still
      offer a simple reliable system, especially for touring cyclist.
    sentences:
      - >-
        This type places the shifters closer to the palm placement, but still
        offer a simple reliable system, especially for touring cyclist.
      - >-
        All square dancers learn standard "definitions" of calls, which they
        recall and use when the caller issues a certain directive.
      - >-
        Mainos-TV operated by leasing atmospheric duration from Yleisradio,
        broadcasting in reserved blocks between Yleisradio's own programming on
        its two channels.
  - source_sentence: >-
      He also played with the Turkish 2nd Division team Pertevniyal, which was
      at the time the farm team of Efes, via a dual license.
    sentences:
      - >-
        The group is still active, producing a monthly action points on the
        women, peace, and authentication blocks affecting countries on Council's
        agenda.
      - >-
        Storage/centre tracks are found in the vicinity of the following
        stations:

        Other song highlights.
      - >-
        He also played with the Turkish 2nd Division team Pertevniyal, which was
        at the time the farm team of Efes, via a two-part authorization.
model-index:
  - name: SentenceTransformer based on google/flan-t5-base
    results:
      - task:
          type: binary-classification
          name: Binary Classification
        dataset:
          name: quora duplicates dev
          type: quora-duplicates-dev
        metrics:
          - type: cosine_accuracy
            value: 0.614
            name: Cosine Accuracy
          - type: cosine_accuracy_threshold
            value: 0.964033842086792
            name: Cosine Accuracy Threshold
          - type: cosine_f1
            value: 0.6810035842293908
            name: Cosine F1
          - type: cosine_f1_threshold
            value: 0.9199645519256592
            name: Cosine F1 Threshold
          - type: cosine_precision
            value: 0.5307262569832403
            name: Cosine Precision
          - type: cosine_recall
            value: 0.95
            name: Cosine Recall
          - type: cosine_ap
            value: 0.6300846409755155
            name: Cosine Ap
          - type: dot_accuracy
            value: 0.53
            name: Dot Accuracy
          - type: dot_accuracy_threshold
            value: 3.121588945388794
            name: Dot Accuracy Threshold
          - type: dot_f1
            value: 0.6666666666666666
            name: Dot F1
          - type: dot_f1_threshold
            value: 1.7629711627960205
            name: Dot F1 Threshold
          - type: dot_precision
            value: 0.5010060362173038
            name: Dot Precision
          - type: dot_recall
            value: 0.996
            name: Dot Recall
          - type: dot_ap
            value: 0.4998742415317971
            name: Dot Ap
          - type: manhattan_accuracy
            value: 0.616
            name: Manhattan Accuracy
          - type: manhattan_accuracy_threshold
            value: 9.075502395629883
            name: Manhattan Accuracy Threshold
          - type: manhattan_f1
            value: 0.6818181818181819
            name: Manhattan F1
          - type: manhattan_f1_threshold
            value: 16.287639617919922
            name: Manhattan F1 Threshold
          - type: manhattan_precision
            value: 0.5286343612334802
            name: Manhattan Precision
          - type: manhattan_recall
            value: 0.96
            name: Manhattan Recall
          - type: manhattan_ap
            value: 0.6295013501048787
            name: Manhattan Ap
          - type: euclidean_accuracy
            value: 0.614
            name: Euclidean Accuracy
          - type: euclidean_accuracy_threshold
            value: 0.41224515438079834
            name: Euclidean Accuracy Threshold
          - type: euclidean_f1
            value: 0.6818181818181819
            name: Euclidean F1
          - type: euclidean_f1_threshold
            value: 0.7474431991577148
            name: Euclidean F1 Threshold
          - type: euclidean_precision
            value: 0.5286343612334802
            name: Euclidean Precision
          - type: euclidean_recall
            value: 0.96
            name: Euclidean Recall
          - type: euclidean_ap
            value: 0.6282929125909658
            name: Euclidean Ap
          - type: max_accuracy
            value: 0.616
            name: Max Accuracy
          - type: max_accuracy_threshold
            value: 9.075502395629883
            name: Max Accuracy Threshold
          - type: max_f1
            value: 0.6818181818181819
            name: Max F1
          - type: max_f1_threshold
            value: 16.287639617919922
            name: Max F1 Threshold
          - type: max_precision
            value: 0.5307262569832403
            name: Max Precision
          - type: max_recall
            value: 0.996
            name: Max Recall
          - type: max_ap
            value: 0.6300846409755155
            name: Max Ap

SentenceTransformer based on google/flan-t5-base

This is a sentence-transformers model finetuned from google/flan-t5-base on the PiC/phrase_similarity dataset. 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: google/flan-t5-base
Maximum Sequence Length: None tokens
Output Dimensionality: 768 tokens
Similarity Function: Cosine Similarity
Training Dataset:
- PiC/phrase_similarity
Language: en

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': None, 'do_lower_case': False}) with Transformer model: T5EncoderModel 
  (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})
)

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("Deehan1866/finetuned-flan-t5-base")
# Run inference
sentences = [
    'He also played with the Turkish 2nd Division team Pertevniyal, which was at the time the farm team of Efes, via a dual license.',
    'He also played with the Turkish 2nd Division team Pertevniyal, which was at the time the farm team of Efes, via a two-part authorization.',
    'Storage/centre tracks are found in the vicinity of the following stations:\nOther song highlights.',
]
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

Binary Classification

Dataset: quora-duplicates-dev
Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.614
cosine_accuracy_threshold	0.964
cosine_f1	0.681
cosine_f1_threshold	0.92
cosine_precision	0.5307
cosine_recall	0.95
cosine_ap	0.6301
dot_accuracy	0.53
dot_accuracy_threshold	3.1216
dot_f1	0.6667
dot_f1_threshold	1.763
dot_precision	0.501
dot_recall	0.996
dot_ap	0.4999
manhattan_accuracy	0.616
manhattan_accuracy_threshold	9.0755
manhattan_f1	0.6818
manhattan_f1_threshold	16.2876
manhattan_precision	0.5286
manhattan_recall	0.96
manhattan_ap	0.6295
euclidean_accuracy	0.614
euclidean_accuracy_threshold	0.4122
euclidean_f1	0.6818
euclidean_f1_threshold	0.7474
euclidean_precision	0.5286
euclidean_recall	0.96
euclidean_ap	0.6283
max_accuracy	0.616
max_accuracy_threshold	9.0755
max_f1	0.6818
max_f1_threshold	16.2876
max_precision	0.5307
max_recall	0.996
max_ap	0.6301

Training Details

Training Dataset

PiC/phrase_similarity

Dataset: PiC/phrase_similarity at fc67ce7
Size: 7,004 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 11 tokens
mean: 28.1 tokens
max: 73 tokens

min: 11 tokens
mean: 28.82 tokens
max: 74 tokens

0: ~48.80%
1: ~51.20%

	sentence1	sentence2	label
type	string	string	int
details	min: 11 tokens mean: 28.1 tokens max: 73 tokens	min: 11 tokens mean: 28.82 tokens max: 74 tokens	0: ~48.80% 1: ~51.20%

Samples:

sentence1	sentence2	label
`newly formed camp is released from the membrane and diffuses across the intracellular space where it serves to activate pka.`	`recently made encampment is released from the membrane and diffuses across the intracellular space where it serves to activate pka.`	`0`
`According to one data, in 1910, on others – in 1915, the mansion became Natalya Dmitriyevna Shchuchkina's property.`	`According to a particular statistic, in 1910, on others – in 1915, the mansion became Natalya Dmitriyevna Shchuchkina's property.`	`1`
`Note that Fact 1 does not assume any particular structure on the set formula_65.`	`Note that Fact 1 does not assume any specific edifice on the set formula_65.`	`0`

Loss: SoftmaxLoss

Evaluation Dataset

PiC/phrase_similarity

Dataset: PiC/phrase_similarity at fc67ce7
Size: 1,000 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 9 tokens
mean: 27.86 tokens
max: 66 tokens

min: 11 tokens
mean: 28.62 tokens
max: 66 tokens

0: ~50.00%
1: ~50.00%

	sentence1	sentence2	label
type	string	string	int
details	min: 9 tokens mean: 27.86 tokens max: 66 tokens	min: 11 tokens mean: 28.62 tokens max: 66 tokens	0: ~50.00% 1: ~50.00%

Samples:

sentence1	sentence2	label
`after theo's apparent death, she decides to leave first colony and ends up traveling with the apostles.`	`after theo's apparent death, she decides to leave original settlement and ends up traveling with the apostles.`	`0`
`The guard assigned to Vivian leaves her to prevent the robbery, allowing her to connect to the bank's network.`	`The guard assigned to Vivian leaves her to prevent the robbery, allowing her to connect to the bank's locations.`	`0`
`Two days later Louis XVI banished Necker by a "lettre de cachet" for his very public exchange of pamphlets.`	`Two days later Louis XVI banished Necker by a "lettre de cachet" for his very free forum of pamphlets.`	`0`

Loss: SoftmaxLoss

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
learning_rate: 2e-05
num_train_epochs: 5
warmup_ratio: 0.1
load_best_model_at_end: True

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
learning_rate: 2e-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: 5
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: False
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: 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
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	quora-duplicates-dev_max_ap
0	0	-	-	0.6114
0.2283	100	-	0.6937	0.6118
0.4566	200	-	0.6934	0.6120
0.6849	300	-	0.6933	0.6118
0.9132	400	-	0.6934	0.6123
1.1416	500	0.6931	0.6933	0.6117
1.3699	600	-	0.6933	0.6117
1.5982	700	-	0.6933	0.6118
1.8265	800	-	0.6933	0.6130
2.0548	900	-	0.6932	0.6130
2.2831	1000	0.6922	0.6931	0.6137
2.5114	1100	-	0.6931	0.6129
2.7397	1200	-	0.6930	0.6143
2.9680	1300	-	0.6928	0.6165
3.1963	1400	-	0.6926	0.6174
3.4247	1500	0.6907	0.6924	0.6193
3.6530	1600	-	0.6920	0.6228
3.8813	1700	-	0.6918	0.6238
4.1096	1800	-	0.6915	0.6256
4.3379	1900	-	0.6912	0.6273
4.5662	2000	0.6888	0.6910	0.6292
4.7945	2100	-	0.6908	0.6301
5.0	2190	-	-	0.6301

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.10.10
Sentence Transformers: 3.0.1
Transformers: 4.42.3
PyTorch: 2.2.1+cu121
Accelerate: 0.32.1
Datasets: 2.20.0
Tokenizers: 0.19.1

Citation

BibTeX

Sentence Transformers and SoftmaxLoss

@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",
}