metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:100
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
base_model: Snowflake/snowflake-arctic-embed-l
widget:
  - source_sentence: >-
      What did the author plan to do with the dark meat and carcass after
      cooking the turkey?
    sentences:
      - >-
        Let’s say a family of four wants to spend only $365 per month on
        groceries, saving them $579 per month over that USDA average family in
        the link above. Investing this savings would compound into about
        $102,483.00 every ten years, which would obviously make a pretty big
        improvement in the financial health of the average young family.

        To hit a monthly grocery spending target like that, you first have to
        understand what you are buying. There are four mouths to feed, each
        consuming three meals a day or 91.25 meals per month. Let’s say they all
        need adult levels of calories, so about 2000 per day.
      - "When you eat beans and rice in the same meal, you’re getting complete protein at virtually no cost. Nuts and especially peanut butter are also a good way to mix high calories with built-in protein. Eggs\_contain the highest quality complete protein of all (6 grams per egg), so I enjoy three of them every day."
      - >-
        Turkey 101 Follow-up

        Thought I’d share how my freezer “spring clean” is going. In an attempt
        to reduce the number of trips to the grocery store in April, I’ve taken
        on the challenge to use up what I have first. Here’s my first attempt at
        staying away from the deli-counter:

        Day 1- After anxiously awaiting the 3 day defrost, ready to cook turkey!
        Easy enough. Since I usually overcook meat (just to make sure it’s
        dead), decided to cook it breast side down; using gravity to my
        advantage, resulting in big, juicy breasts (just like my hubby likes).
        Save dark meat for later. Freeze some white meat, slice some for
        sandwiches, make broth from carcass.
  - source_sentence: >-
      What are the benefits of using whole oils in your diet according to the
      context?
    sentences:
      - >-
        What to Eat

        Finally, the fun part! As the wise people of India have proven beyond
        all other cultures*, amazing food is all about preparation and spices,
        rather than starting with costly ingredients. Once you know which
        ingredients make good staples, you can easily poke around on the
        Internet or in any cookbook to find an infinite number of good recipes
        that use them.

        At the simplest “bachelor” level, you’ve got recipes like:

        Fancy home fries:
      - "Aha.. now things are sounding much better. Although not all of the foods above cost less than $1 per meal, they can certainly average out to less than that, depending on how you combine them.\_And when planning your menu to meet a certain budget, averaging out is exactly your goal. You still want to be able to eat apples, organic chicken breast, or whatever your heart desires. You just have to not eat entirely\_those most expensive foods.\nAnd remember, this $1.00 target is just something I picked out of a hat for an example – you’re allowed to spend whatever works for you."
      - >-
        Whole oils are the ultimate example. They are packed with tasty,
        slow-metabolizing calories, extremely good for you, and easy to mix into
        your diet. Using olive oil as an example, you can one third of a day
        worth of calories for 57 cents. Every time you dump these oils into a
        frying pan, or mix them into a recipe or a salad dressing, you’re
        lowering your food cost – the oil provides calories that your body might
        otherwise get from cans of Coke, Filet Mignon, or Burger King dollar
        menu burgers.
  - source_sentence: >-
      What ingredients did the "Master Mix" consist of, and how was it used in
      cooking?
    sentences:
      - >-
        Day 4- Morph yesterdays’ meal into a turkey pot pie. Thankfully, pie
        crust does not require yeast….I think. Decide to skip the 99 cent
        pre-packaged spice mix, and make my own taco seasoning?! I don’t have
        any maltodextrin, modified corn starch, autolyzed yeast extract, or
        caramel color (sulfites) in my cupboard; so hope it turns out okay. Cook
        up the remaining meat for turkey tacos, and freeze half for later.

        Day 5- Enjoy eating leftovers.
      - >-
        This is a fantastic article.  I’m generally responsible for our family’s
        grocery shopping since I do the dinner cooking.  Our budget is $185 for
        a family of four per two weeks (two boys are almost 4 and 16 months). 
        Some two-weeks are tight, but it’s been worthwhile for our bottom line
        to keep the budget set.  We also budget $20 for restaurants per 2
        weeks.  Yes, I know we can’t go out on that, but if we save it up, we
        can go out once a month or so, or order pizza one week, or some
        combination.  I’m sure our budget will increase when the boys get older,
        but by then, we should be bringing in more money, so we plan on being
        able to absorb the increase.  Eating healthy and abundantly doesn’t have
        to be expensive, but it does require work and
      - >-
        When I was growing up, my parents had 9 mouths to feed, and I remember
        my mom making something called a “Master Mix”.  It was basically a
        biscuit mix with the butter mixed in already, which she kept in a
        4-liter ice cream pail.  She’d use it to make pizza dough (among other
        things),  and she’d top it with canned tomato soup (still condensed),
        shredded carrots and broccoli and cheddar cheese.  My siblings and I
        have confessed an occasional desire to eat it again, although I don’t
        know I’d ever try it out on my own kids.


        Reply







        Diane

        April 9, 2020, 11:30 pm
  - source_sentence: >-
      What changes were made to the homeowners insurance policy to achieve a
      $600 reduction?
    sentences:
      - >-
        And contrary to the 1990s low-fat-diet fad, the human body loves oil.
        It’s yummy, clean-burning, good for a giant range of body functions, and
        it is satisfying to eat too. I eat a fairly high-fat/low-carb diet these
        days, yet I’m leaner than ever, because the oily food doesn’t cause
        spikes of fake appetite like bread does. I’ve even been known to bring
        containers of herb-infused olive oil on road trips, supplementing every
        meal with this supercharger nutrient, especially when it’s time for an
        extreme hike or a high-energy work day.

        See Article: The Amazing Waist-Slimming, Wallet-Fattening Nutrient
      - >-
        First thing- reduced insurance by $600 with increasing the homeowners
        deductible from $500 to $1000, and switching providers. Be warned- was
        not informed about the “unannounced 3rd party” that would be knocking on
        my door, as well as the additional cost to reappraise some items- but
        still overall a reduction. Second- dropped the gym membership
        ($131/month). Now don’t have to feel guilty about not going. Enjoy the
        outdoors more anyhow. Third- scaled back on vacation. I’m actually “on
        vacation” everyday, as even with all the expenses, we’re at FI.
      - >-
        Reply





        beachmama

        January 31, 2017, 11:39 am



        As a 25+ year veg, 12 year vegan, I’ve always supplemented b-12. After
        getting blood work done I found I was critically low in D3. Turns out
        it’s not just because I’m a woman over 50 (now 61) and through
        menopause, or that I’ve been veg for over half my life, I’m fit and walk
        the beach 20 miles a week so getting sun isn’t enough even in
        California. Apparently most people are D3 deficient but never know until
        they become symptomatic or have a blood test. I recommend you get a
        simple test to check on b-12 and d3 just to make sure you’re in good
        shape. And you are SO right about protein . . . Westerners eat FAR too
        much protein ; )


        Reply













        riley

        March 29, 2012, 7:07 am
  - source_sentence: >-
      What additional ingredients are suggested to increase protein content in
      the context?
    sentences:
      - "Those are just two simple recipes. The key to frugal eating is to have at least ten good things you know how to make.\nThere are many chefs among the readers. Maybe we will get to hear some of their best low-cost and easy-to-make creations in the comments section below?\nFurther Reading:\nGrocery Shopping with your Middle Finger – an old MMM classic on this same topic, where I first started thinking about cost per calorie. But there I\_ was dealing with food stockups and sales rather than thinking of it on a per-meal or per-month basis.\n* According to the strong opinion of my own taste buds"
      - >-
        Thanks for this timely article! In the midst of the March Challenge; was
        trying to determine the next item to tackle- and groceries was it! How’d
        you know it was $1000? Hmmm….psychic.

        I FINALLY updated all the spending on Quicken last month to make myself
        stare it in the face. No surprises; not ugly, but not very pretty
        either. The most valuable outcome of the exercise was showing my husband
        that his hard efforts are appreciated, and I’m stepping up!
      - >-
        cocoa and maybe some ground flax or whatever is lying around) for an
        extra 40 grams of protein.
pipeline_tag: sentence-similarity
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
model-index:
  - name: SentenceTransformer based on Snowflake/snowflake-arctic-embed-l
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: cosine_accuracy@1
            value: 0.7582417582417582
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 0.9120879120879121
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 0.945054945054945
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 0.9725274725274725
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 0.7582417582417582
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.304029304029304
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.18901098901098898
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.09725274725274723
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 0.7582417582417582
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 0.9120879120879121
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 0.945054945054945
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 0.9725274725274725
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 0.870936179086928
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 0.837580673294959
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 0.8395868579934513
            name: Cosine Map@100

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

This is a sentence-transformers model finetuned from Snowflake/snowflake-arctic-embed-l. 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: Snowflake/snowflake-arctic-embed-l
Maximum Sequence Length: 512 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': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 1024, '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 additional ingredients are suggested to increase protein content in the context?',
    'cocoa and maybe some ground flax or whatever is lying around) for an extra 40 grams of protein.',
    'Thanks for this timely article! In the midst of the March Challenge; was trying to determine the next item to tackle- and groceries was it! How’d you know it was $1000? Hmmm….psychic.\nI FINALLY updated all the spending on Quicken last month to make myself stare it in the face. No surprises; not ugly, but not very pretty either. The most valuable outcome of the exercise was showing my husband that his hard efforts are appreciated, and I’m stepping up!',
]
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]

Evaluation

Metrics

Information Retrieval

Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7582
cosine_accuracy@3	0.9121
cosine_accuracy@5	0.9451
cosine_accuracy@10	0.9725
cosine_precision@1	0.7582
cosine_precision@3	0.304
cosine_precision@5	0.189
cosine_precision@10	0.0973
cosine_recall@1	0.7582
cosine_recall@3	0.9121
cosine_recall@5	0.9451
cosine_recall@10	0.9725
cosine_ndcg@10	0.8709
cosine_mrr@10	0.8376
cosine_map@100	0.8396

Training Details

Training Dataset

Unnamed Dataset

Size: 100 training samples
Columns: sentence_0 and sentence_1
Approximate statistics based on the first 100 samples:
sentence_0 sentence_1
type string string
details
min: 12 tokens
mean: 17.78 tokens
max: 27 tokens

min: 7 tokens
mean: 125.38 tokens
max: 195 tokens

	sentence_0	sentence_1
type	string	string
details	min: 12 tokens mean: 17.78 tokens max: 27 tokens	min: 7 tokens mean: 125.38 tokens max: 195 tokens

Samples:

sentence_0	sentence_1
`What strategies might be suggested for reducing a $1000 grocery bill?`	`Killing your $1000 Grocery Bill Home Media Contact Email RSS Start Here About Random MMM Recommends Forum MMM Classics Mr. Money Mustache View: Fancy Magazine`
`When was the article "Killing your $1000 Grocery Bill" published?`	`Killing your $1000 Grocery Bill Home Media Contact Email RSS Start Here About Random MMM Recommends Forum MMM Classics Mr. Money Mustache View: Fancy Magazine`
`What type of event was the narrator attending where they enjoyed a potluck buffet?`	A few years ago, I was at a party eating some amazing food at the potluck buffet. In my area, there seems to be a friendly competition among the thirtysomething outdoorsy tech worker crowd, of trying to out-chef each other. It’s a contest I heartily approve of and I am happy to be both an underdog competitor and a judge. Anyway, the topic turned to how good we have it in our lives, with such plentiful food that we can afford to spend hours combining exotic ingredients just for the sake of overfilling our bellies. “Yeah… I know it’s a bit over the top”, I said, “but we probably spend 80 bucks a week on good groceries. I think it’s worth it if you can afford it”.

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: 10
per_device_eval_batch_size: 10
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: 10
per_device_eval_batch_size: 10
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: 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	cosine_ndcg@10
1.0	10	0.8684
2.0	20	0.8698
3.0	30	0.8699
4.0	40	0.8706
5.0	50	0.8709

Framework Versions

Python: 3.13.1
Sentence Transformers: 3.4.1
Transformers: 4.48.3
PyTorch: 2.6.0
Accelerate: 1.3.0
Datasets: 3.2.0
Tokenizers: 0.21.0

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}
}