notebook_test / README.md

electroglyph

Upload model

a89bfe1 verified 19 days ago

preview code

raw

history blame contribute delete

14 kB

metadata

language:
  - en
tags:
  - unsloth
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - dense
  - generated_from_trainer
  - dataset_size:106628
  - loss:MultipleNegativesRankingLoss
base_model: Alibaba-NLP/gte-modernbert-base
widget:
  - source_sentence: ace-v
    sentences:
      - >-
        The floor plan was drafted at 1/4 inch scale where each quarter inch
        equals one foot.
      - Fingerprint examiners follow the ACE-V methodology for identification.
      - Most modern streaming services offer content in 1080p full HD quality.
  - source_sentence: adult learner
    sentences:
      - The adult learner brings valuable life experience to the classroom.
      - Accounts payable represents money owed to suppliers and vendors.
      - The inspection confirmed all above grade work met code requirements.
  - source_sentence: 1/4 inch scale
    sentences:
      - Precise adjustments require accurate action gauge readings.
      - The quality inspector identified adhesion failure in the sample.
      - >-
        The architect created drawings at 1/4 inch scale for the client
        presentation.
  - source_sentence: acrylic paint
    sentences:
      - Artists prefer acrylic paint for its fast drying time.
      - The company reported strong adjusted EBITDA growth this quarter.
      - The clinic specializes in adolescent health services.
  - source_sentence: adult learning
    sentences:
      - Solar developers calculate AEP, or annual energy production.
      - The course was designed using adult learning best practices.
      - >-
        The wizard cast Abi-Dalzim's horrid wilting, draining moisture from
        enemies.
datasets:
  - electroglyph/technical
pipeline_tag: sentence-similarity
library_name: sentence-transformers

SentenceTransformer

This model was finetuned with Unsloth.

based on Alibaba-NLP/gte-modernbert-base

This is a sentence-transformers model finetuned from Alibaba-NLP/gte-modernbert-base on the technical 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: Alibaba-NLP/gte-modernbert-base
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 dimensions
Similarity Function: Cosine Similarity
Training Dataset:
- technical
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': 512, 'do_lower_case': False, 'architecture': 'PeftModelForFeatureExtraction'})
  (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})
)

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 = [
    'adult learning',
    'The course was designed using adult learning best practices.',
    'Solar developers calculate AEP, or annual energy production.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.7228, 0.1468],
#         [0.7228, 1.0000, 0.1683],
#         [0.1468, 0.1683, 1.0000]])

Training Details

Training Dataset

technical

Dataset: technical at 05eeb90
Size: 106,628 training samples
Columns: anchor and positive
Approximate statistics based on the first 1000 samples:
anchor positive
type string string
details
min: 3 tokens
mean: 4.93 tokens
max: 12 tokens

min: 9 tokens
mean: 13.75 tokens
max: 25 tokens

	anchor	positive
type	string	string
details	min: 3 tokens mean: 4.93 tokens max: 12 tokens	min: 9 tokens mean: 13.75 tokens max: 25 tokens

Samples:

anchor	positive
`.308`	`The .308 Winchester is a popular rifle cartridge used for hunting and target shooting.`
`.308`	`Many precision rifles are chambered in .308 for its excellent long-range accuracy.`
`.308`	`The sniper selected a .308 caliber round for the mission.`

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: 333
learning_rate: 3e-05
num_train_epochs: 5
lr_scheduler_type: constant_with_warmup
warmup_steps: 100
fp16: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
prediction_loss_only: True
per_device_train_batch_size: 333
per_device_eval_batch_size: 8
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: 3e-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: constant_with_warmup
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
warmup_steps: 100
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
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}
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
project: huggingface
trackio_space_id: trackio
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: no
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: True
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss
0.1558	50	3.4086
0.3115	100	3.3329
0.4673	150	3.2148
0.6231	200	2.9797
0.7788	250	2.7541
0.9346	300	2.5277
1.0903	350	2.3069
1.2461	400	2.1593
1.4019	450	2.0781
1.5576	500	1.9385
1.7134	550	1.9052
1.8692	600	1.8768
2.0249	650	1.8272
2.1807	700	1.7906
2.3364	750	1.7607
2.4922	800	1.7375
2.6480	850	1.6952
2.8037	900	1.6664
2.9595	950	1.6216
3.1153	1000	1.5601
3.2710	1050	1.571
3.4268	1100	1.5735
3.5826	1150	1.5455
3.7383	1200	1.5577
3.8941	1250	1.5426
4.0498	1300	1.5276
4.2056	1350	1.5178
4.3614	1400	1.4611
4.5171	1450	1.4822
4.6729	1500	1.4987
4.8287	1550	1.4507
4.9844	1600	1.4501

Framework Versions

Python: 3.12.12
Sentence Transformers: 5.2.0
Transformers: 4.57.3
PyTorch: 2.9.0+cu126
Accelerate: 1.12.0
Datasets: 4.3.0
Tokenizers: 0.22.1

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}
}