metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:268861
  - loss:MultipleNegativesRankingLoss
base_model: Qwen/Qwen3-0.6B-Base
widget:
  - source_sentence: >-
      There are seven thieves. They stole diamonds from a diamond merchant and
      ran away. While running, night sets in and they decide to rest in the
      jungle.

      When everybody was sleeping, two of them woke up and decided to divide the
      diamonds equally among themselves. But when they divided the diamonds
      equally, one diamond is left.

      So they woke up the 3rd thief and tried to divide the diamonds equally
      again but still one diamond was left. Then they woke up the 4th thief to
      divide the diamonds equally again, and again one diamond was left. This
      happened with the 5th and 6th thief – one diamond was still left.

      Finally, they woke up the 7th thief and this time the diamonds were
      divided equally.

      How many diamonds did they steal in total?
    sentences:
      - ''''
      - ''''
      - e
  - source_sentence: >-
      praveen starts business with rs . 3220 and after 5 months , hari joins
      with praveen as his partner . after a year , the profit is divided in the
      ratio 2 : 3 . what is hari ’ s contribution in the capital ?
    sentences:
      - s
      - '5'
      - '['
  - source_sentence: |-
      Which of the following is material of choice in class V
      cavity with abfraction?
    sentences:
      - '['
      - t
      - G
  - source_sentence: >-
      A right circular cylinder has a height of 25 and a radius of 5. A
      rectangular solid with a height of 15 and a square base, is placed in the
      cylinder such that each of the corners of the solid is tangent to the
      cylinder wall. Liquid is then poured into the cylinder such that it
      reaches the rim. What is the volume of the liquid?
    sentences:
      - '5'
      - '['
      - '2'
  - source_sentence: Cerebral angiography was performed by -
    sentences:
      - S
      - t
      - '2'
pipeline_tag: sentence-similarity
library_name: sentence-transformers

SentenceTransformer based on Qwen/Qwen3-0.6B-Base

This is a sentence-transformers model finetuned from Qwen/Qwen3-0.6B-Base. 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: Qwen/Qwen3-0.6B-Base
Maximum Sequence Length: 128 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': 128, 'do_lower_case': False}) with Transformer model: Qwen3Model 
  (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 = [
    'Cerebral angiography was performed by -',
    'S',
    '2',
]
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: 268,861 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: 5 tokens
mean: 48.3 tokens
max: 128 tokens

min: 0 tokens
mean: 0.97 tokens
max: 1 tokens

	sentence_0	sentence_1
type	string	string
details	min: 5 tokens mean: 48.3 tokens max: 128 tokens	min: 0 tokens mean: 0.97 tokens max: 1 tokens

Samples:

sentence_0	sentence_1
`A 1200 m long train crosses a tree in 120 sec, how much time will I take to pass a platform 1100 m long?`	`'`
`What is the opposite of rarefaction zones, where air molecules in waves are loosely packed?`	`[`
`if w is 40 percent less than e , e is 40 percent less than y , and z is 46 percent less than y , then z is greater than w by what percent of w ?`	`%`

Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 64
per_device_eval_batch_size: 64
num_train_epochs: 4
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: 64
per_device_eval_batch_size: 64
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: 4
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}
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
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	Training Loss
0.1190	500	4.0939
0.2380	1000	3.7716
0.3571	1500	0.0
0.4761	2000	0.0
0.5951	2500	0.0
0.7141	3000	0.0
0.8331	3500	0.0
0.9522	4000	0.0
1.0712	4500	0.0
1.1902	5000	0.0
1.3092	5500	0.0
1.4282	6000	0.0
1.5473	6500	0.0
1.6663	7000	0.0
1.7853	7500	0.0
1.9043	8000	0.0
2.0233	8500	0.0
2.1423	9000	0.0
2.2614	9500	0.0
2.3804	10000	0.0
2.4994	10500	0.0
2.6184	11000	0.0
2.7374	11500	0.0
2.8565	12000	0.0
2.9755	12500	0.0
3.0945	13000	0.0
3.2135	13500	0.0
3.3325	14000	0.0
3.4516	14500	0.0
3.5706	15000	0.0
3.6896	15500	0.0
3.8086	16000	0.0
3.9276	16500	0.0

Framework Versions

Python: 3.11.13
Sentence Transformers: 4.1.0
Transformers: 4.52.4
PyTorch: 2.6.0+cu124
Accelerate: 1.7.0
Datasets: 3.6.0
Tokenizers: 0.21.1

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}
}