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 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': 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("CatkinChen/sentence-transformers_all-MiniLM-L6-v2_retrieval_finetuned_2025-03-29_23-44-08")
# Run inference
sentences = [
    'What is the name of the magical candy that explodes?',
    'Book: 3, Chapter: 10\nPassage: Harry reached the door at the top of the stairs, slipped through it, and found himself behind the counter of Honeydukes - he ducked, crept sideways, and then straightened up. Honeydukes was so crowded with Hogwarts students that no one looked twice at Harry. He edged among them, looking around, and suppressed a laugh as he imagined the look that would spread over Dudley\'s piggy face if he could see where Harry was now. There were shelves upon shelves of the most succulent-looking sweets imaginable. Creamy chunks of nougat, shimmering pink squares of coconut ice, fat, honey-colored toffees; hundreds of different kinds of chocolate in neat rows; there was a large barrel of Every Flavor Beans, and another of Fizzing Whizbees, the levitating sherbert balls that Ron had mentioned; along yet another wall were "Special Effects" sweets: Drooble\'s Best Blowing Gum (which filled a room with bluebell-colored bubbles that refused to pop for days), the strange, splintery Toothflossing Stringmints, tiny black Pepper Imps ("breathe fire for your friends! "), Ice Mice ("hear your teeth chatter and squeak! "), peppermint creams shaped like toads ("hop realistically in the stomach! "), fragile sugar-spun quills, and exploding bonbons. Harry squeezed himself through a crowd of sixth years and saw a sign hanging in the farthest corner of the shop (UNUSUAL TASTES).',
    'Book: 2, Chapter: 8\nPassage: But I mean to say - look at the fellow -"\n"I think," said Harry hurriedly, at a meaningful look from Nick, "Nick\'s very - frightening and - er -"\n"Ha!" yelled Sir Patrick\'s head. "Bet he asked you to say that!" "If I could have everyone\'s attention, it\'s time for my speech!" said Nearly Headless Nick loudly, striding toward the podium and climbing into an icy blue spotlight. "My late lamented lords, ladies, and gentlemen, it is my great sorrow ..."\nBut nobody heard much more. Sir Patrick and the rest of the Headless Hunt had just started a game of Head Hockey and the crowd were turning to watch. Nearly Headless Nick tried vainly to recapture his audience, but gave up as Sir Patrick\'s head went sailing past him to loud cheers. Harry was very cold by now, not to mention hungry.',
]
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

Triplet

Dataset: test-eval
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.8

Training Details

Training Dataset

Unnamed Dataset

Size: 309 training samples
Columns: sentence_0, sentence_1, and sentence_2
Approximate statistics based on the first 309 samples:
sentence_0 sentence_1 sentence_2
type string string string
details
min: 9 tokens
mean: 17.74 tokens
max: 53 tokens
min: 46 tokens
mean: 187.05 tokens
max: 256 tokens
min: 40 tokens
mean: 163.65 tokens
max: 256 tokens

Samples:

sentence_0	sentence_1	sentence_2
`Who is the founder of Gryffindor house?`	Book: 2, Chapter: 9[object Object]Passage: He peered at Hermione as though he had never seen a student properly before. "However, the legend of which you speak is such a very sensational, even ludicrous tale -"[object Object]But the whole class was now hanging on Professor Binns's every word. He looked dimly at them all, every face turned to his. Harry could tell he was completely thrown by such an unusual show of interest. "Oh, very well," he said slowly. "Let me see ... the Chamber of Secrets ...[object Object]"You all know, of course, that Hogwarts was founded over a thousand years ago - the precise date is uncertain - by the four greatest witches and wizards of the age. The four school Houses are named after them: Godric Gryffindor, Helga Hufflepuff, Rowena Ravenclaw, and Salazar Slytherin. They built this castle together, far from prying Muggle eyes, for it was an age when magic was feared by common people, and witches and wizards suffered much persecution."	Book: 3, Chapter: 15[object Object]Passage: "Down here -"[object Object]Harry looked quickly back out of the window. Crookshanks and the dog had vanished. Harry climbed onto the windowsill to look right down into the shadows of the castle, but they weren't there. Where had they gone? A loud snore told him Ron had fallen asleep again. Harry and the rest of the Gryffindor team entered the Great Hall the next day to enormous applause. Harry couldn't help grinning broadly as he saw that both the Ravenclaw and Hufflepuff tables were applauding them too.
`What magical creature does Hagrid keep in his hut in Harry's third year?`	Book: 3, Chapter: 11[object Object]Passage: choked Hagrid, wiping his eyes on his sleeve. "They've got it in fer interestin' creatures!" A sudden sound from the corner of Hagrid's cabin made Harry, Ron, and Hermione whip around. Buckbeak the hippogriff was lying in the corner, chomping on something that was oozing blood all over the floor. "I couldn' leave him tied up out there in the snow!" choked Hagrid. "All on his own! At Christmas." Harry, Ron, and Hermione looked at one another. They had never seen eye to eye with Hagrid about what he called "interesting creatures" and other people called "terrifying monsters." On the other hand, there didn't seem to be any particular harm in Buckbeak.	Book: 7, Chapter: 15[object Object]Passage: "So we still haven't got any food." "Shut up, Ron," snapped Hermione. "Harry, what happened? Why do you think you couldn't make your Patronus? You managed perfectly yesterday!" "I don't know." He sat low in one of Perkins's old armchairs, feeling more humiliated by the moment. He was afraid that something had gone wrong inside him. Yesterday seemed a long time ago: Today he might have been thirteen years old again, the only one who collapsed on the Hogwarts Express.
`What magical plant traps Harry, Ron, and Hermione in their first year?`	Book: 1, Chapter: 16[object Object]Passage: Hermione had managed to free herself before the plant got a firm grip on her. Now she watched in horror as the two boys fought to pull the plant off them, but the more they strained against it, the tighter and faster the plant wound around them. "Stop moving!" Hermione ordered them. "I know what this is - it's Devil's Snare!" "Oh, I'm so glad we know what it's called, that's a great help," snarled Ron, leaning back, trying to stop the plant from curling around his neck.	Book: 7, Chapter: 35[object Object]Passage: He lay facedown, listening to the silence. He was perfectly alone. Nobody was watching. Nobody else was there. He was not perfectly sure that he was there himself. A long time later, or maybe no time at all, it came to him that he must exist, must be more than disembodied thought, because he was lying, definitely lying, on some surface. Therefore he had a sense of touch, and the thing against which he lay existed too. Almost as soon as he had reached this conclusion, Harry became conscious that he was naked. Convinced as he was of his total solitude, this did not concern him, but it did intrigue him slightly. He wondered whether, as he could feel, he would be able to see.

Loss: TripletLoss with these parameters:
json
1{ 2 "distance_metric": "TripletDistanceMetric.COSINE", 3 "triplet_margin": 0.5 4}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 4
per_device_eval_batch_size: 4
num_train_epochs: 1
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 4
per_device_eval_batch_size: 4
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: 1
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}
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

Epoch	Step	test-eval_cosine_accuracy
0.1282	10	0.75
0.2564	20	0.7643
0.3846	30	0.8000
0.5128	40	0.8071
0.6410	50	0.8000
0.7692	60	0.8000
0.8974	70	0.8000
1.0	78	0.8000
-1	-1	0.8000

Framework Versions

Python: 3.12.2
Sentence Transformers: 3.4.1
Transformers: 4.50.0
PyTorch: 2.4.1
Accelerate: 1.4.0
Datasets: 2.19.2
Tokenizers: 0.21.0

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",
}

TripletLoss

bibtex

@misc{hermans2017defense,
    title={In Defense of the Triplet Loss for Person Re-Identification},
    author={Alexander Hermans and Lucas Beyer and Bastian Leibe},
    year={2017},
    eprint={1703.07737},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

	sentence_0	sentence_1	sentence_2
type	string	string	string
details	min: 9 tokens mean: 17.74 tokens max: 53 tokens	min: 46 tokens mean: 187.05 tokens max: 256 tokens	min: 40 tokens mean: 163.65 tokens max: 256 tokens