This is a sentence-transformers model finetuned from thenlper/gte-small on the json 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: thenlper/gte-small
Maximum Sequence Length: 512 tokens
Output Dimensionality: 384 dimensions
Similarity Function: Cosine Similarity
Training Dataset:
- json

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': False, 'architecture': 'BertModel'})
  (1): Pooling({'word_embedding_dimension': 384, '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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("1tunadogan/gte-small-financial-matryoshka")
# Run inference
sentences = [
    "How much did GM Financial's primary source of cash from finance charge income increase in 2023 compared to the previous year?",
    'In the year ended December 31, 2023, Net cash provided by operating activities increased primarily due to an increase in finance charge income of $1.7 billion.',
    "A corporate entity referred to as a management services organization (MSO) provides various management services and keeps the physician entity 'friendly' through a stock transfer restriction agreement and/or other relationships. The fees under the management services arrangement must comply with state fee splitting laws, which in some states may prohibit percentage-based fees.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.6872, 0.2882],
#         [0.6872, 1.0000, 0.3045],
#         [0.2882, 0.3045, 1.0000]])

Evaluation

Metrics

Information Retrieval

Dataset: dim_384
Evaluated with InformationRetrievalEvaluator with these parameters:
json
1{ 2 "truncate_dim": 384 3}

Metric	Value
cosine_accuracy@1	0.6914
cosine_accuracy@3	0.82
cosine_accuracy@5	0.8614
cosine_accuracy@10	0.8957
cosine_precision@1	0.6914
cosine_precision@3	0.2733
cosine_precision@5	0.1723
cosine_precision@10	0.0896
cosine_recall@1	0.6914
cosine_recall@3	0.82
cosine_recall@5	0.8614
cosine_recall@10	0.8957
cosine_ndcg@10	0.7945
cosine_mrr@10	0.7618
cosine_map@100	0.7656

Information Retrieval

Dataset: dim_256
Evaluated with InformationRetrievalEvaluator with these parameters:
json
1{ 2 "truncate_dim": 256 3}

Metric	Value
cosine_accuracy@1	0.68
cosine_accuracy@3	0.8157
cosine_accuracy@5	0.8557
cosine_accuracy@10	0.8914
cosine_precision@1	0.68
cosine_precision@3	0.2719
cosine_precision@5	0.1711
cosine_precision@10	0.0891
cosine_recall@1	0.68
cosine_recall@3	0.8157
cosine_recall@5	0.8557
cosine_recall@10	0.8914
cosine_ndcg@10	0.7879
cosine_mrr@10	0.7545
cosine_map@100	0.7583

Information Retrieval

Dataset: dim_128
Evaluated with InformationRetrievalEvaluator with these parameters:
json
1{ 2 "truncate_dim": 128 3}

Metric	Value
cosine_accuracy@1	0.6743
cosine_accuracy@3	0.8043
cosine_accuracy@5	0.8386
cosine_accuracy@10	0.8814
cosine_precision@1	0.6743
cosine_precision@3	0.2681
cosine_precision@5	0.1677
cosine_precision@10	0.0881
cosine_recall@1	0.6743
cosine_recall@3	0.8043
cosine_recall@5	0.8386
cosine_recall@10	0.8814
cosine_ndcg@10	0.778
cosine_mrr@10	0.7448
cosine_map@100	0.7491

Information Retrieval

Dataset: dim_64
Evaluated with InformationRetrievalEvaluator with these parameters:
json
1{ 2 "truncate_dim": 64 3}

Metric	Value
cosine_accuracy@1	0.65
cosine_accuracy@3	0.7729
cosine_accuracy@5	0.8243
cosine_accuracy@10	0.8643
cosine_precision@1	0.65
cosine_precision@3	0.2576
cosine_precision@5	0.1649
cosine_precision@10	0.0864
cosine_recall@1	0.65
cosine_recall@3	0.7729
cosine_recall@5	0.8243
cosine_recall@10	0.8643
cosine_ndcg@10	0.7562
cosine_mrr@10	0.7217
cosine_map@100	0.7263

Training Details

Training Dataset

json

Dataset: json
Size: 6,300 training samples
Columns: anchor and positive
Approximate statistics based on the first 1000 samples:
anchor positive
type string string
details
min: 7 tokens
mean: 20.44 tokens
max: 45 tokens
min: 8 tokens
mean: 45.16 tokens
max: 512 tokens

Samples:

anchor	positive
`What was the amount of cash generated from operations by the company in fiscal year 2023?`	`Highlights during fiscal year 2023 include the following: We generated $18,085 million of cash from operations.`
`How much were unrealized losses on U.S. government and agency securities for those held for 12 months or greater as of June 30, 2023?`	`U.S. government and agency securities \| $ \| 7,950 \| \| $ \| (336 \| ) \| $ \| 45,273 \| $ \| (3,534 \| ) \| $ \| 53,223 \| $ \| (3,870 \| )`
`How is the impairment of assets assessed for projects still under development?`	`For assets under development, assets are grouped and assessed for impairment by estimating the undiscounted cash flows, which include remaining construction costs, over the asset's remaining useful life. If cash flows do not exceed the carrying amount, impairment based on fair value versus carrying value is considered.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 384, 5 256, 6 128, 7 64 8 ], 9 "matryoshka_weights": [ 10 1.0, 11 1.0, 12 1.0, 13 1.0 14 ], 15 "n_dims_per_step": -1 16}

Evaluation Dataset

json

Dataset: json
Size: 700 evaluation samples
Columns: anchor and positive
Approximate statistics based on the first 700 samples:
anchor positive
type string string
details
min: 9 tokens
mean: 20.9 tokens
max: 51 tokens
min: 6 tokens
mean: 48.7 tokens
max: 512 tokens

Samples:

anchor	positive
`How much were the company's debt obligations as of December 31, 2023?`	`The company's debt obligations as of December 31, 2023, totaled $2,299,887 thousand.`
`What are the specific structures and legal considerations for a management services organization (MSO) in relation to its relationship with physician owners?`	`A corporate entity referred to as a management services organization (MSO) provides various management services and keeps the physician entity 'friendly' through a stock transfer restriction agreement and/or other relationships. The fees under the management services arrangement must comply with state fee splitting laws, which in some states may prohibit percentage-based fees.`
`Where does Eli Lilly and Company manufacture and distribute its products?`	`We manufacture and distribute our products through facilities in the United States (U.S.), including Puerto Rico, and in Europe and Asia.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 384, 5 256, 6 128, 7 64 8 ], 9 "matryoshka_weights": [ 10 1.0, 11 1.0, 12 1.0, 13 1.0 14 ], 15 "n_dims_per_step": -1 16}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
gradient_accumulation_steps: 32
learning_rate: 2e-05
num_train_epochs: 4
lr_scheduler_type: cosine
warmup_ratio: 0.1
tf32: False
load_best_model_at_end: True
optim: adamw_torch
dataloader_pin_memory: False
gradient_checkpointing: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: epoch
prediction_loss_only: True
per_device_train_batch_size: 8
per_device_eval_batch_size: 8
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 32
eval_accumulation_steps: None
torch_empty_cache_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: 4
max_steps: -1
lr_scheduler_type: cosine
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: False
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}
parallelism_config: 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: False
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
hub_revision: None
gradient_checkpointing: True
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
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss	dim_384_cosine_ndcg@10	dim_256_cosine_ndcg@10	dim_128_cosine_ndcg@10	dim_64_cosine_ndcg@10
0.4061	10	1.8735	-	-	-	-	-
0.8122	20	0.8955	-	-	-	-	-
1.0	25	-	0.2409	0.7822	0.7763	0.7623	0.7490
1.2030	30	0.3892	-	-	-	-	-
1.6091	40	0.2253	-	-	-	-	-
2.0	50	0.1488	0.1453	0.7919	0.7801	0.7701	0.7479
2.4061	60	0.1474	-	-	-	-	-
2.8122	70	0.1287	-	-	-	-	-
3.0	75	-	0.1280	0.7951	0.7883	0.7762	0.7562
3.2030	80	0.0948	-	-	-	-	-
3.6091	90	0.0952	-	-	-	-	-
4.0	100	0.0974	0.1285	0.7945	0.7879	0.778	0.7562

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.12.11
Sentence Transformers: 5.1.0
Transformers: 4.56.1
PyTorch: 2.8.0
Accelerate: 1.10.1
Datasets: 4.1.0
Tokenizers: 0.22.0

Citation

BibTeX

Sentence Transformers

bibtex

@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",
}

MatryoshkaLoss

bibtex

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

MultipleNegativesRankingLoss

bibtex

@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}
}

	anchor	positive
type	string	string
details	min: 7 tokens mean: 20.44 tokens max: 45 tokens	min: 8 tokens mean: 45.16 tokens max: 512 tokens