SentenceTransformer based on Alibaba-NLP/gte-Qwen2-1.5B-instruct

This is a sentence-transformers model finetuned from Alibaba-NLP/gte-Qwen2-1.5B-instruct. It maps sentences & paragraphs to a 1536-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: Alibaba-NLP/gte-Qwen2-1.5B-instruct
Maximum Sequence Length: 512 tokens
Output Dimensionality: 1536 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': False, 'architecture': 'Qwen2Model'})
  (1): Pooling({'word_embedding_dimension': 1536, '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})
)

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 = [
    'General supervision means that the physician need not be physically present at the patient\'s place of residence when the service is performed; however, the service must be performed under his or her overall supervision and control The physician orders the service(s) to be performed, and contact is maintained between the nurse or other employee and the physician, e.g., the employee contacts the physician directly if additional instructions are needed, and the physician must retain professional responsibility for the service All other "incident to" requirements must be met (see §§60-60.4). 3 The services are included in the physician\'s/clinic\'s bill, and the physician or clinic has incurred an expense for them (see §60.2). 4 The services of the paramedical are required for the patient\'s care; that is, they are reasonable and necessary as defined in the Medicare Benefit Policy Manual, Chapter 16, "General Exclusions from Coverage," §20. 5 When the service can be furnished by an HHA in the local area, it cannot be covered when furnished by a physician/clinic to a homebound patient under this provision, except as described in §60.4.C.',
    'General supervision means that the physician need not be physically present at the patient\'s place of residence when the service is performed; however, the service must be performed under his or her overall supervision and control The physician orders the service(s) to be performed, and contact is maintained between the nurse or other employee and the physician, e.g., the employee contacts the physician directly if additional instructions are needed, and the physician must retain professional responsibility for the service All other "incident to" requirements must be met (see §§60-60.4). 3 The services are included in the physician\'s/clinic\'s bill, and the physician or clinic has incurred an expense for them (see §60.2). 4 The services of the paramedical are required for the patient\'s care; that is, they are reasonable and necessary as defined in the Medicare Benefit Policy Manual, Chapter 16, "General Exclusions from Coverage," §20. 5 When the service can be furnished by an HHA in the local area, it cannot be covered when furnished by a physician/clinic to a homebound patient under this provision, except as described in §60.4.C.',
    'Implementation: 11-01-24) Transmit the CMS-2591 to CO via PC or terminal Use instructions in the CROWD User Guide available via the CMS Enterprise Portal The report is due as soon as possible after the end of the reporting month but no later than the 15th of the month following the end of the reporting month.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1536]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[ 1.0000,  1.0000, -0.1049],
#         [ 1.0000,  1.0000, -0.1049],
#         [-0.1049, -0.1049,  1.0000]])

Training Details

Training Dataset

Unnamed Dataset

Size: 60,906 training samples
Columns: sentence_0 and sentence_1
Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1
type string string
details
min: 32 tokens
mean: 145.93 tokens
max: 512 tokens

min: 32 tokens
mean: 145.93 tokens
max: 512 tokens

	sentence_0	sentence_1
type	string	string
details	min: 32 tokens mean: 145.93 tokens max: 512 tokens	min: 32 tokens mean: 145.93 tokens max: 512 tokens

Samples:

sentence_0	sentence_1
The preadmission screening in the patient's IRF medical record serves as the primary documentation by the IRF clinical staff of the patient's status prior to admission and of the specific reasons that led the IRF clinical staff to conclude that the IRF admission would be reasonable and necessary As such, IRFs must make this documentation detailed and comprehensive In accordance with 42 CFR § 412.622(a)(4)(i)(B) the preadmission screening documentation must indicate the patient's prior level of function (prior to the event or condition that led to the patient's need for intensive rehabilitation therapy), expected level of improvement, and the expected length of time necessary to achieve that level of improvement	The preadmission screening in the patient's IRF medical record serves as the primary documentation by the IRF clinical staff of the patient's status prior to admission and of the specific reasons that led the IRF clinical staff to conclude that the IRF admission would be reasonable and necessary As such, IRFs must make this documentation detailed and comprehensive In accordance with 42 CFR § 412.622(a)(4)(i)(B) the preadmission screening documentation must indicate the patient's prior level of function (prior to the event or condition that led to the patient's need for intensive rehabilitation therapy), expected level of improvement, and the expected length of time necessary to achieve that level of improvement
and (C) An attestation that the component organization will prominently post notification on its Web site and publish in any promotional materials for dissemination to providers, a summary of the information that is required by paragraph (c)(4)(i)(A) of this section. (ii) Comply with the following requirements during its period of listing: (A) The component organization may not share staff with its parent organization(s). (B) The component organization may enter into a written agreement pursuant to paragraph (c)(3) but such agreements are limited to units or individuals of the parent organization(s) whose responsibilities do not involve the activities specified in the restrictions in paragraph (a)(2)(ii) of this section	and (C) An attestation that the component organization will prominently post notification on its Web site and publish in any promotional materials for dissemination to providers, a summary of the information that is required by paragraph (c)(4)(i)(A) of this section. (ii) Comply with the following requirements during its period of listing: (A) The component organization may not share staff with its parent organization(s). (B) The component organization may enter into a written agreement pursuant to paragraph (c)(3) but such agreements are limited to units or individuals of the parent organization(s) whose responsibilities do not involve the activities specified in the restrictions in paragraph (a)(2)(ii) of this section
Review of the person-centered active treatment plan. (d) The CMHC interdisciplinary treatment team must review, revise, and document the individualized active treatment plan as frequently as the client's condition requires, but no less frequently than every 30-calendar day A revised active treatment plan must include information from the client's initial evaluation and comprehensive assessments, the client's progress toward outcomes and goals specified in the active treatment plan, and changes in the client's goals The CMHC must also meet partial hospitalization program requirements specified under § 424.24(e) of this chapter or intensive outpatient service requirements as specified under § 424.24(d) of this chapter, as applicable, if such services are included in the active treatment plan	Review of the person-centered active treatment plan. (d) The CMHC interdisciplinary treatment team must review, revise, and document the individualized active treatment plan as frequently as the client's condition requires, but no less frequently than every 30-calendar day A revised active treatment plan must include information from the client's initial evaluation and comprehensive assessments, the client's progress toward outcomes and goals specified in the active treatment plan, and changes in the client's goals The CMHC must also meet partial hospitalization program requirements specified under § 424.24(e) of this chapter or intensive outpatient service requirements as specified under § 424.24(d) of this chapter, as applicable, if such services are included in the active treatment plan

Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 16
per_device_eval_batch_size: 16
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
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
num_train_epochs: 3
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: 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: 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}
parallelism_config: None
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch_fused
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
hub_revision: None
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
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: round_robin
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss
0.1313	500	0.173
0.2627	1000	0.1505
0.3940	1500	0.1613
0.5253	2000	0.1568
0.6567	2500	0.1677
0.7880	3000	0.1611
0.9194	3500	0.1571
1.0507	4000	0.1688
1.1820	4500	0.1682
1.3134	5000	0.1609
1.4447	5500	0.1621
1.5760	6000	0.1528
1.7074	6500	0.1576
1.8387	7000	0.1581
1.9701	7500	0.1591
2.1014	8000	0.1479
2.2327	8500	0.1623
2.3641	9000	0.1572
2.4954	9500	0.1577
2.6267	10000	0.158
2.7581	10500	0.16
2.8894	11000	0.1693

Framework Versions

Python: 3.12.6
Sentence Transformers: 5.2.0
Transformers: 4.56.0
PyTorch: 2.8.0+cu129
Accelerate: 1.10.1
Datasets: 4.4.1
Tokenizers: 0.22.0

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}
}