SentenceTransformer based on answerdotai/ModernBERT-large

This is a sentence-transformers model finetuned from answerdotai/ModernBERT-large. It maps sentences & paragraphs to a 1024-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: answerdotai/ModernBERT-large
Maximum Sequence Length: 512 tokens
Output Dimensionality: 1024 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}) with Transformer model: ModernBertModel 
  (1): Pooling({'word_embedding_dimension': 1024, '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'what is a tensilon universal testing instrument',
    "Universal Material Testing Instrument. The TENSILON RTF is our newest universal testing machine offering innovative measuring possibilities, based on A&D's newly-developed and extensive technological knowledge.The RTF Series is a world-class Class 0.5 testing machine.Having improved the overall design and structure of the machine, we achieved a very strong load frame stiffness enabling super-high accuracy in measurement.he RTF Series is a world-class Class 0.5 testing machine. Having improved the overall design and structure of the machine, we achieved a very strong load frame stiffness enabling super-high accuracy in measurement.",
    "The McDonald Patent Universal String Tension Calculator (MPUSTC) is a handy calculator to figure string tensions in steel-string instruments. If you plug in your scale length, string gauges and tuning, it will give you a readout of the tension on each of the strings. This is useful when you're trying to fine-tune a set of custom gauges, or when you're working out how far you can push a drop tuning before it becomes unmanageable.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

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

Training Details

Training Dataset

Unnamed Dataset

Size: 499,184 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: 9.07 tokens max: 21 tokens	min: 17 tokens mean: 80.89 tokens max: 254 tokens	min: 20 tokens mean: 79.05 tokens max: 226 tokens

Samples:

sentence_0	sentence_1	sentence_2
`what is a dependent person`	`1. depending on a person or thing for aid, support, life, etc. 2. (postpositive; foll by on or upon) influenced or conditioned (by); contingent (on) 3. subordinate; subject: a dependent prince. 4. obsolete hanging down.`	Dependent personality disorder (DPD) is one of the most frequently diagnosed personality disorders. It occurs equally in men and women, usually becoming apparent in young adulthood or later as important adult relationships form. People with DPD become emotionally dependent on other people and spend great effort trying to please others. People with DPD tend to display needy, passive, and clinging behavior, and have a fear of separation. Other common characteristics of this personality disorder include:
`what is the hat trick in hockey`	`Definition of hat trick. 1 1 : the retiring of three batsmen with three consecutive balls by a bowler in cricket. 2 2 : the scoring of three goals in one game (as of hockey or soccer) by a single player. 3 3 : a series of three victories, successes, or related accomplishments scored a hat trick when her three best steers corralled top honors â People.`	`Hat trick was first recorded in print in the 1870s, but has since been widened to apply to any sport in which the person competing carries off some feat three times in quick succession, such as scoring three goals in one game of soccer.`
`what is an egalitarian`	`An egalitarian is defined as a person who believes all people were created equal and should be treated equal. An example of an egalitarian is a person who fights for civil rights, like Martin Luther King Jr.`	About Egalitarian Companies. In the tradition hierarchical corporate structure, each employee operates under a specific job description. Each employee also reports to a superior who monitors his progress and issues instructions. Egalitarian-style companies eliminate most of this structure. Employees in an egalitarian company have general job descriptions, rather than specific ones. Instead of reporting to a superior, all employees in an egalitarian company work collaboratively on tasks and behave as equals.

Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 32
per_device_eval_batch_size: 32
num_train_epochs: 10
fp16: True
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: 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: 10
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
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: batch_sampler
multi_dataset_batch_sampler: round_robin

Training Logs