khoj-ai/timely-arctic-medium

Technical Report and Model Pipeline

To access our technical report and model pipeline scripts visit our github

SentenceTransformer based on Snowflake/snowflake-arctic-embed-m

This is a sentence-transformers model finetuned from Snowflake/snowflake-arctic-embed-m. 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: Snowflake/snowflake-arctic-embed-m
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 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': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (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})
  (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.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("sentence_transformers_model_id")
# Run inference
sentences = [
    'Represent this sentence for searching relevant passages: Witki, Warmian-Masurian Voivodeship 2040 Oct 12',
    'Witki () is a village in the administrative district of Gmina Bartoszyce, within Bartoszyce County, Warmian-Masurian Voivodeship, in northern Poland, close to the border with the Kaliningrad Oblast of Russia. It lies approximately east of Bartoszyce and north-east of the regional capital Olsztyn. References Witki 12/10/2040\n',
    '12-21-2046 This is a list of electoral results for the Electoral district of Irwin in Western Australian state elections. Members for Irwin Election results Elections in the 1940s  Preferences were not distributed.  Preferences were not distributed. Elections in the 1930s  Preferences were not distributed. Elections in the 1920s Elections in the 1910s Elections in the 1900s Elections in the 1890s References Western Australian state electoral results by district\n',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

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

Training Details

Training Dataset

Unnamed Dataset

• Size: 55,744 training samples
• Columns: anchors and positive
• Approximate statistics based on the first 1000 samples:

  	anchors	positive
  type	string	string
  details	min: 14 tokens mean: 20.33 tokens max: 33 tokens	min: 17 tokens mean: 48.58 tokens max: 75 tokens

• Samples:

  anchors	positive
  Represent this sentence for searching relevant passages: are bugs attracted to citronella November 10?	Citronella is naturally occurring oil that repels insects. ... “Citronella oil is repellent to mosquitoes to a degree, but the amount being put out by a candle isn't going to be very effective,” Eric Hoffer, president of Hoffer Pest, told TODAY Home. 11/10
  Represent this sentence for searching relevant passages: are bugs attracted to citronella 11/10/09?	Citronella is naturally occurring oil that repels insects. ... “Citronella oil is repellent to mosquitoes to a degree, but the amount being put out by a candle isn't going to be very effective,” Eric Hoffer, president of Hoffer Pest, told TODAY Home. 11/10/09
  Represent this sentence for searching relevant passages: are bugs attracted to citronella Jan 15?	Citronella is naturally occurring oil that repels insects. ... “Citronella oil is repellent to mosquitoes to a degree, but the amount being put out by a candle isn't going to be very effective,” Eric Hoffer, president of Hoffer Pest, told TODAY Home. 01/15

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 1,000 evaluation samples
• Columns: anchors and positive
• Approximate statistics based on the first 1000 samples:

  	anchors	positive
  type	string	string
  details	min: 12 tokens mean: 21.57 tokens max: 43 tokens	min: 8 tokens mean: 66.44 tokens max: 512 tokens

• Samples:

  anchors	positive
  Represent this sentence for searching relevant passages: Identify a Psychopath 3/28	Psychopathy is a personality construct consisting of a cluster of characteristics used by mental health professionals to describe someone who is charming, manipulative, emotionally ruthless and potentially criminal. 03/28
  Represent this sentence for searching relevant passages: what is dangerous high blood pressure in pregnancy?	A blood pressure that is greater than 130/90 mm Hg or that is 15 degrees higher on the top number from where you started before pregnancy may be cause for concern. High blood pressure during pregnancy is defined as 140 mm Hg or higher systolic, with diastolic 90 mm Hg or higher.
  Represent this sentence for searching relevant passages: Be a Better Cheerleader June 22	What do you think when you think of a good cheerleader? Tight with motions? Can hold a stunt? Well, it's not just that. You need to be fit in 3 categories: mental/emotional health, social health, and physical health. 06/22

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 128
• per_device_eval_batch_size: 64
• learning_rate: 1.5e-05
• weight_decay: 0.01
• num_train_epochs: 1
• warmup_ratio: 0.1
• warmup_steps: 400
• bf16: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 128
• 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: 1.5e-05
• weight_decay: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 1
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• warmup_steps: 400
• 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: True
• 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}
• 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
• eval_use_gather_object: False
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss
0.0023	1	2.4713	-
0.0229	10	2.4907	-
0.0459	20	2.4574	-
0.0688	30	2.4861	-
0.0917	40	2.4612	-
0.1147	50	2.4353	-
0.1376	60	2.3967	-
0.1606	70	2.3609	-
0.1835	80	2.3079	-
0.2064	90	2.1928	-
0.2294	100	2.1581	-
0.2523	110	2.0822	-
0.2752	120	1.9739	-
0.2982	130	1.8393	-
0.3211	140	1.7397	-
0.3440	150	1.5249	-
0.3670	160	1.4281	-
0.3899	170	1.3197	-
0.4128	180	1.211	-
0.4358	190	1.1086	-
0.4587	200	0.9598	0.2301
0.4817	210	1.0904	-
0.5046	220	0.9813	-
0.5275	230	1.1148	-
0.5505	240	1.2813	-
0.5734	250	1.2259	-
0.5963	260	1.221	-
0.6193	270	1.1547	-
0.6422	280	1.1286	-
0.6651	290	0.9932	-
0.6881	300	0.978	-
0.7110	310	0.9505	-
0.7339	320	0.8731	-
0.7569	330	0.824	-
0.7798	340	0.8979	-
0.8028	350	1.756	-
0.8257	360	1.6785	-
0.8486	370	1.5944	-
0.8716	380	1.5417	-
0.8945	390	1.4788	-
0.9174	400	0.9873	0.0695
0.9404	410	0.1664	-
0.9633	420	0.1336	-
0.9862	430	0.1193	-

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.43.4
• PyTorch: 2.4.0+cu121
• Accelerate: 0.33.0
• Datasets: 2.20.0
• Tokenizers: 0.19.1

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}
}