metadata

base_model: Snowflake/snowflake-arctic-embed-m
library_name: sentence-transformers
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@3
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_precision@1
  - cosine_precision@3
  - cosine_precision@5
  - cosine_precision@10
  - cosine_recall@1
  - cosine_recall@3
  - cosine_recall@5
  - cosine_recall@10
  - cosine_ndcg@10
  - cosine_mrr@10
  - cosine_map@100
  - dot_accuracy@1
  - dot_accuracy@3
  - dot_accuracy@5
  - dot_accuracy@10
  - dot_precision@1
  - dot_precision@3
  - dot_precision@5
  - dot_precision@10
  - dot_recall@1
  - dot_recall@3
  - dot_recall@5
  - dot_recall@10
  - dot_ndcg@10
  - dot_mrr@10
  - dot_map@100
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:600
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
widget:
  - source_sentence: What types of additional risks might future updates incorporate?
    sentences:
      - >-
        Inaccuracies in these labels can impact the “stability” or robustness of
        these benchmarks, which many GAI practitioners consider during the model
        selection process.
      - >-
        For example, when prompted to generate images of CEOs, doctors, lawyers,
        and judges, current text-to-image models underrepresent women and/or
        racial minorities , and people with disabilities .
      - >-
        Future updates may incorporate additional risks or provide further
        details on the risks identiﬁed below.
  - source_sentence: >-
      What are some potential consequences of the abuse and misuse of AI systems
      by humans?
    sentences:
      - >-
        Even when trained on “clean” data, increasingly capable GAI models can
        synthesize or produce synthetic NCII and CSAM.
      - >-
        3 the abuse, misuse, and unsafe repurposing by humans (adversarial or
        not ), and others result from interactions between a human and an AI
        system.
      - >-
        Energy and carbon emissions vary based on what is being done with the
        GAI model (i.e., pre -training, ﬁne -tuning, inference), the modality of
        the content , hardware used, and type of task or application .
  - source_sentence: What types of digital content can be included in GAI?
    sentences:
      - >-
        Errors in t hird-party GAI components can also have downstream impacts
        on accuracy and robustness .
      - >-
        In direct prompt injections, attackers might craft malicious prompts and
        input them directly to a GAI system , with a variety of downstream
        negative consequences to interconnected systems.
      - >-
        This can include images, videos, audio, text, and other digital
        content.” While not all GAI is derived from foundation models, for
        purposes of this document, GAI generally refers to generative foundation
        models .
  - source_sentence: >-
      What are the implications of harmful bias and homogenization in relation
      to stereotypical content?
    sentences:
      - >-
        These risks provide a lens through which organizations can frame and
        execute risk management eﬀorts.
      - >-
        13 • Not every suggested action appl ies to every AI Actor14 or is
        relevant to every AI Actor Task .
      - >-
        The spread of denigrating or stereotypical content can also further
        exacerbate representational harms (see Harmful Bias and Homogenization
        below).
  - source_sentence: >-
      What are the inventory exemptions defined in organizational policies for
      GAI systems embedded into application software?
    sentences:
      - >-
        Methods for creating smaller versions of train ed models, such as model
        distillation or compression, could reduce environmental impacts at
        inference time, but training and tuning such models may still contribute
        to their environmental impacts .
      - >-
        For example, predictive inferences made by GAI models based on PII or
        protected attributes c an contribute to adverse decisions , leading to
        representational or allocative harms to individuals or groups (see
        Harmful Bias and Homogenization below).
      - >-
        Information Security GV-1.6-002 Deﬁne any inventory exemptions in
        organizational policies for GAI systems embedded into application
        software .
model-index:
  - name: SentenceTransformer based on Snowflake/snowflake-arctic-embed-m
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: cosine_accuracy@1
            value: 0.9
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.98
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.99
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 1
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.9
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.3266666666666667
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.19799999999999998
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.09999999999999998
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.9
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.98
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.99
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 1
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.9563669441556807
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.9417619047619047
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.9417619047619047
            name: Cosine Map@100
          - type: dot_accuracy@1
            value: 0.9
            name: Dot Accuracy@1
          - type: dot_accuracy@3
            value: 0.98
            name: Dot Accuracy@3
          - type: dot_accuracy@5
            value: 0.99
            name: Dot Accuracy@5
          - type: dot_accuracy@10
            value: 1
            name: Dot Accuracy@10
          - type: dot_precision@1
            value: 0.9
            name: Dot Precision@1
          - type: dot_precision@3
            value: 0.3266666666666667
            name: Dot Precision@3
          - type: dot_precision@5
            value: 0.19799999999999998
            name: Dot Precision@5
          - type: dot_precision@10
            value: 0.09999999999999998
            name: Dot Precision@10
          - type: dot_recall@1
            value: 0.9
            name: Dot Recall@1
          - type: dot_recall@3
            value: 0.98
            name: Dot Recall@3
          - type: dot_recall@5
            value: 0.99
            name: Dot Recall@5
          - type: dot_recall@10
            value: 1
            name: Dot Recall@10
          - type: dot_ndcg@10
            value: 0.9563669441556807
            name: Dot Ndcg@10
          - type: dot_mrr@10
            value: 0.9417619047619047
            name: Dot Mrr@10
          - type: dot_map@100
            value: 0.9417619047619047
            name: Dot Map@100

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 = [
    'What are the inventory exemptions defined in organizational policies for GAI systems embedded into application software?',
    'Information Security GV-1.6-002 Deﬁne any inventory exemptions in organizational policies for GAI systems embedded into application software .',
    'For example, predictive inferences made by GAI models based on PII or protected attributes c an contribute to adverse decisions , leading to representational or allocative harms to individuals or groups (see Harmful Bias and Homogenization below).',
]
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]

Evaluation

Metrics

Information Retrieval

Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.9
cosine_accuracy@3	0.98
cosine_accuracy@5	0.99
cosine_accuracy@10	1.0
cosine_precision@1	0.9
cosine_precision@3	0.3267
cosine_precision@5	0.198
cosine_precision@10	0.1
cosine_recall@1	0.9
cosine_recall@3	0.98
cosine_recall@5	0.99
cosine_recall@10	1.0
cosine_ndcg@10	0.9564
cosine_mrr@10	0.9418
cosine_map@100	0.9418
dot_accuracy@1	0.9
dot_accuracy@3	0.98
dot_accuracy@5	0.99
dot_accuracy@10	1.0
dot_precision@1	0.9
dot_precision@3	0.3267
dot_precision@5	0.198
dot_precision@10	0.1
dot_recall@1	0.9
dot_recall@3	0.98
dot_recall@5	0.99
dot_recall@10	1.0
dot_ndcg@10	0.9564
dot_mrr@10	0.9418
dot_map@100	0.9418

Training Details

Training Dataset

Unnamed Dataset

Size: 600 training samples
Columns: sentence_0 and sentence_1
Approximate statistics based on the first 600 samples:
sentence_0 sentence_1
type string string
details
min: 7 tokens
mean: 18.93 tokens
max: 33 tokens

min: 4 tokens
mean: 43.35 tokens
max: 165 tokens

	sentence_0	sentence_1
type	string	string
details	min: 7 tokens mean: 18.93 tokens max: 33 tokens	min: 4 tokens mean: 43.35 tokens max: 165 tokens

Samples:

sentence_0	sentence_1
`What are indirect prompt injections and how can they exploit vulnerabilities?`	`Security researchers have already demonstrated how indirect prompt injections can exploit vulnerabilities by steal ing proprietary data or running malicious code remotely on a machine.`
`What potential consequences can arise from exploiting vulnerabilities through indirect prompt injections?`	`Security researchers have already demonstrated how indirect prompt injections can exploit vulnerabilities by steal ing proprietary data or running malicious code remotely on a machine.`
`What factors might organizations consider when tailoring their measurement of GAI risks?`	`Organizations may choose to tailor how they measure GAI risks based on these characteristics .`

Loss: MatryoshkaLoss with these parameters:

{
    "loss": "MultipleNegativesRankingLoss",
    "matryoshka_dims": [
        768,
        512,
        256,
        128,
        64
    ],
    "matryoshka_weights": [
        1,
        1,
        1,
        1,
        1
    ],
    "n_dims_per_step": -1
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 20
per_device_eval_batch_size: 20
num_train_epochs: 5
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: 20
per_device_eval_batch_size: 20
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: 5
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}
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: batch_sampler
multi_dataset_batch_sampler: round_robin

Training Logs

Epoch	Step	cosine_map@100
1.0	30	0.9216
1.6667	50	0.9292
2.0	60	0.9361
3.0	90	0.9418

Framework Versions

Python: 3.11.9
Sentence Transformers: 3.1.1
Transformers: 4.45.0
PyTorch: 2.4.1+cu121
Accelerate: 0.34.2
Datasets: 3.0.1
Tokenizers: 0.20.0

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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

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}
}