This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2. 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: sentence-transformers/all-MiniLM-L6-v2
Maximum Sequence Length: 256 tokens
Output Dimensionality: 384 tokens
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': 256, 'do_lower_case': False}) with Transformer model: 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("mazej/all-MiniLM-L6-v2_MultipleNegativesRankingLoss_b64_141306_fine-tuned")
# Run inference
sentences = [
    'When should the word masters or bachelors be capitalized with apostrophe s?',
    'When someone has more than one master’s degree, should these be described as have several _masters’_ degrees or several _master’s_ degrees? In other words, which of these two applies: (singular) a master’s degree > (plural) several master’s degrees (singular) a master’s degree > (plural) several masters’ degrees Please note that this is a different question from what this has been marked as a duplicate of. That addresses the question for the singular. This question is about the plural case. There is definitely an apostrophe, but the question is where it should go.',
    'It doesnt seem to be consistent how many skills I must level up in order to gain a character level. It also doesnt seem to be a whole number, since I currently have about 2 pixels of bar left to get from level 3 to level 4. What is the formula like for determining when the character gains a level, based on what skills level?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

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

Evaluation

Metrics

Information Retrieval

Dataset: ir_eval
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@10	0.7102
cosine_precision@10	0.0929
cosine_recall@10	0.6518
cosine_ndcg@10	0.5426
cosine_mrr@10	0.5355
cosine_map@100	0.5012
dot_accuracy@10	0.7102
dot_precision@10	0.0929
dot_recall@10	0.6518
dot_ndcg@10	0.5426
dot_mrr@10	0.5355
dot_map@100	0.5012

Training Details

Training Dataset

Unnamed Dataset

Size: 1,836 training samples
Columns: query and document
Approximate statistics based on the first 1000 samples:
query document
type string string
details
min: 4 tokens
mean: 12.87 tokens
max: 44 tokens
min: 10 tokens
mean: 127.62 tokens
max: 256 tokens

Samples:

query	document
`Shortest correct sentence in English- use of contractions`	Is there some rule against ending a sentence with the contraction its? Sometimes its fine to use a contraction at the end of a sentence: If youre thinking about starting a land war in Asia, just dont! (Sounds fine!) Other times it sounds wrong: Mr Vizzini is smarter than Im. (You must say, ... than I am.) I am unaware of any rule that prescribes when it is okay to use a contraction at the end, and when it is not (or maybe Im ascribing too much importance to the position of the contraction over some more significant attribute?). In any case, if there is a rule, I have no idea what it's.
`Asking question about position of a person in a list`	`I want to ask a question so that I can get the answer which gives the position of the President. So the answer I want to get is: > Barack Hussein Obama is the 44th President of the United States. or > 44th. or simply > 44. How can I properly ask this question?`
`Can NPCs drown?`	`If you have Waterbreathing, but your follower does not, will they follow you into deep water and stay down long enough to drown themselves? I may play-test this myself in the next few days, but I was wondering if anyone else might already know the answer. In any case, I figured it would be good to have this question here for anyone else that might be interested.`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim" 4}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
per_device_train_batch_size: 64
per_device_eval_batch_size: 64
gradient_accumulation_steps: 2
warmup_ratio: 0.1
fp16: True
load_best_model_at_end: 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: 64
per_device_eval_batch_size: 64
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 2
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: 1.0
num_train_epochs: 3
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: 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: 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
use_liger_kernel: False
eval_use_gather_object: False
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	ir_eval_cosine_map@100
0	0	0.5080
1.0345	15	0.5012
2.0345	30	0.5008
2.8276	42	0.5012

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.8.10
Sentence Transformers: 3.2.1
Transformers: 4.45.2
PyTorch: 2.0.0+cu117
Accelerate: 1.0.1
Datasets: 3.1.0
Tokenizers: 0.20.3

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",
}

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}
}

	query	document
type	string	string
details	min: 4 tokens mean: 12.87 tokens max: 44 tokens	min: 10 tokens mean: 127.62 tokens max: 256 tokens