Hugging Face's logo

tomaarsen
/
inference-free-splade-bert-tiny-nq-3e-6-lambda-corpus

Feature Extraction
sentence-transformers
Safetensors
English
sparse-encoder
sparse
asymmetric
inference-free
splade
Generated from Trainer
dataset_size:99000
loss:SpladeLoss
Eval Results
Model card Files
xet
Community
inference-free-splade-bert-tiny-nq-3e-6-lambda-corpus / README.md
tomaarsen's picture
tomaarsen HF Staff
Add new SparseEncoder model
7630c70 verified
preview code
|
raw
history blame contribute delete
48.6 kB
 ---
language:
- en
license: apache-2.0
tags:
- sentence-transformers
- sparse-encoder
- sparse
- asymmetric
- inference-free
- splade
- generated_from_trainer
- dataset_size:99000
- loss:SpladeLoss
widget:
- source_sentence: where is the tiber river located in italy
sentences:
- Sales taxes in British Columbia On 1 July 2010, the PST and GST were combined
into the Harmonized Sales Tax (HST) levied according to the provisions of the
GST. The conversion to HST was controversial. Popular opposition led to a referendum
on the tax system, the first such referendum in the Commonwealth of Nations, resulting
in the province reverting to the former PST/GST model on 1 April 2013.
- 'Tiber The Tiber (/ˈtaɪbər/, Latin: Tiberis,[1] Italian: Tevere [ˈteːvere])[2]
is the third-longest river in Italy, rising in the Apennine Mountains in Emilia-Romagna
and flowing 406 kilometres (252 mi) through Tuscany, Umbria and Lazio, where it
is joined by the river Aniene, to the Tyrrhenian Sea, between Ostia and Fiumicino.[3]
It drains a basin estimated at 17,375 square kilometres (6,709 sq mi). The river
has achieved lasting fame as the main watercourse of the city of Rome, founded
on its eastern banks.'
- 'Water in California California''s limited water supply comes from two main sources:
surface water, or water that travels or gathers on the ground, like rivers, streams,
and lakes; and groundwater, which is water that is pumped out from the ground.
California has also begun producing a small amount of desalinated water, water
that was once sea water, but has been purified.'
- source_sentence: what kind of car does jay gatsby drive
sentences:
- Jay Gatsby At the Buchanan home, Jordan Baker, Nick, Jay, and the Buchanans decide
to visit New York City. Tom borrows Gatsby's yellow Rolls Royce to drive up to
the city. On the way to New York City, Tom makes a detour at a gas station in
"the Valley of Ashes", a run-down part of Long Island. The owner, George Wilson,
shares his concern that his wife, Myrtle, may be having an affair. This unnerves
Tom, who has been having an affair with Myrtle, and he leaves in a hurry.
- 'Panama Canal The Panama Canal (Spanish: Canal de Panamá) is an artificial 77 km
(48 mi) waterway in Panama that connects the Atlantic Ocean with the Pacific Ocean.
The canal cuts across the Isthmus of Panama and is a conduit for maritime trade.
Canal locks are at each end to lift ships up to Gatun Lake, an artificial lake
created to reduce the amount of excavation work required for the canal, 26 m (85
ft) above sea level, and then lower the ships at the other end. The original locks
are 34 m (110 ft) wide. A third, wider lane of locks was constructed between September
2007 and May 2016. The expanded canal began commercial operation on June 26, 2016.
The new locks allow transit of larger, post-Panamax ships, capable of handling
more cargo.[1]'
- Solar maximum Predictions of a future maximum's timing and strength are very difficult;
predictions vary widely. There was a solar maximum in 2000. In 2006 NASA initially
expected a solar maximum in 2010 or 2011, and thought that it could be the strongest
since 1958.[3] However, the solar maximum was not declared to have occurred until
2014, and even then was ranked among the weakest on record.[4]
- source_sentence: who sings if i can dream about you
sentences:
- Wesley Jonathan Wesley Jonathan Waples (born October 18, 1978), known professionally
as Wesley Jonathan, is an American actor. He is best known for his starring role
as Jamal Grant on the NBC Saturday morning comedy-drama series City Guys, Sweetness
in the 2005 film Roll Bounce, as well as Burrell "Stamps" Ballentine on TV Land's
The Soul Man.
- I Can Dream About You "I Can Dream About You" is a song performed by American
singer Dan Hartman on the soundtrack album of the film Streets of Fire. Released
in 1984 as a single from the soundtrack, and included on Hartman's album I Can
Dream About You, it reached number 6 on the Billboard Hot 100.[1]
- Blood is thicker than water In modern society, the proverb "blood is thicker than
water" is used to imply that family relationships are always more important than
friends.
- source_sentence: who did jesse palmer end up with on the bachelor
sentences:
- Jesse Palmer In 2004, Palmer was the first professional athlete to appear on The
Bachelor television program and the first non-American bachelor, in which he was
given his choice of eligible single women. He eventually selected Jessica Bowlin,
but their courtship lasted for only a few months after the end of the show.[19][20]
- Wave base In seawater, the water particles are moved in a circular orbital motion
when a wave passes. The radius of the circle of motion for any given water molecule
decreases exponentially with increasing depth. The wave base, which is the depth
of influence of a water wave, is about half the wavelength.
- Do You Remember the First Time? (The Vampire Diaries) Elena, after everyone continues
to convince her that she had once loved damon decides to run through the magic
free, mystic falls border. So she does, and she gets glimpses of her and Damon
but never fully remembers yet that she loves him. Damon pulls her back across
the line and she asks about a kiss in the rain. He continues to try to get her
to remember.
- source_sentence: when did the american civil rights movement end
sentences:
- 'A Sunday Afternoon on the Island of La Grande Jatte A Sunday Afternoon on the
Island of La Grande Jatte (French: Un dimanche après-midi à l''Île de la Grande
Jatte) painted in 1884, is one of Georges Seurat''s most famous works. It is a
leading example of pointillist technique, executed on a large canvas. Seurat''s
composition includes a number of Parisians at a park on the banks of the River
Seine.'
- Paleolithic Paleolithic humans made tools of stone, bone, and wood.[23] The early
paleolithic hominins, Australopithecus, were the first users of stone tools. Excavations
in Gona, Ethiopia have produced thousands of artifacts, and through radioisotopic
dating and magnetostratigraphy, the sites can be firmly dated to 2.6 million
years ago. Evidence shows these early hominins intentionally selected raw materials
with good flaking qualities and chose appropriate sized stones for their needs
to produce sharp-edged tools for cutting.[29]
- African-American civil rights movement (1954–1968) The Civil Rights Movement (also
known as the American civil rights movement, African-American civil rights movement,
and other terms,[b]) was a human rights movement from 1954–1968 that encompassed
strategies, groups, and social movements to accomplish its goal of ending legalized
racial segregation and discrimination laws in the United States. The movement
secured the legal recognition and federal protection of black Americans in the
United States Constitution and federal law.
datasets:
- sentence-transformers/natural-questions
pipeline_tag: feature-extraction
library_name: sentence-transformers
metrics:
- 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
co2_eq_emissions:
emissions: 11.776380098641885
energy_consumed: 0.030296679972425883
source: codecarbon
training_type: fine-tuning
on_cloud: false
cpu_model: 13th Gen Intel(R) Core(TM) i7-13700K
ram_total_size: 31.777088165283203
hours_used: 0.096
hardware_used: 1 x NVIDIA GeForce RTX 3090
model-index:
- name: Inference-free SPLADE BERT-tiny trained on Natural-Questions tuples
results:
- task:
type: sparse-information-retrieval
name: Sparse Information Retrieval
dataset:
name: NanoMSMARCO
type: NanoMSMARCO
metrics:
- type: dot_accuracy@1
value: 0.28
name: Dot Accuracy@1
- type: dot_accuracy@3
value: 0.54
name: Dot Accuracy@3
- type: dot_accuracy@5
value: 0.68
name: Dot Accuracy@5
- type: dot_accuracy@10
value: 0.74
name: Dot Accuracy@10
- type: dot_precision@1
value: 0.28
name: Dot Precision@1
- type: dot_precision@3
value: 0.18
name: Dot Precision@3
- type: dot_precision@5
value: 0.136
name: Dot Precision@5
- type: dot_precision@10
value: 0.07400000000000001
name: Dot Precision@10
- type: dot_recall@1
value: 0.28
name: Dot Recall@1
- type: dot_recall@3
value: 0.54
name: Dot Recall@3
- type: dot_recall@5
value: 0.68
name: Dot Recall@5
- type: dot_recall@10
value: 0.74
name: Dot Recall@10
- type: dot_ndcg@10
value: 0.5066725139399298
name: Dot Ndcg@10
- type: dot_mrr@10
value: 0.4317460317460317
name: Dot Mrr@10
- type: dot_map@100
value: 0.4432974611015074
name: Dot Map@100
- task:
type: sparse-information-retrieval
name: Sparse Information Retrieval
dataset:
name: NanoNFCorpus
type: NanoNFCorpus
metrics:
- type: dot_accuracy@1
value: 0.44
name: Dot Accuracy@1
- type: dot_accuracy@3
value: 0.58
name: Dot Accuracy@3
- type: dot_accuracy@5
value: 0.58
name: Dot Accuracy@5
- type: dot_accuracy@10
value: 0.64
name: Dot Accuracy@10
- type: dot_precision@1
value: 0.44
name: Dot Precision@1
- type: dot_precision@3
value: 0.3866666666666667
name: Dot Precision@3
- type: dot_precision@5
value: 0.29600000000000004
name: Dot Precision@5
- type: dot_precision@10
value: 0.248
name: Dot Precision@10
- type: dot_recall@1
value: 0.043253729866814
name: Dot Recall@1
- type: dot_recall@3
value: 0.07701448892020092
name: Dot Recall@3
- type: dot_recall@5
value: 0.0882103437254049
name: Dot Recall@5
- type: dot_recall@10
value: 0.11441879984163104
name: Dot Recall@10
- type: dot_ndcg@10
value: 0.3142746286394966
name: Dot Ndcg@10
- type: dot_mrr@10
value: 0.5100555555555555
name: Dot Mrr@10
- type: dot_map@100
value: 0.13631231221886872
name: Dot Map@100
- task:
type: sparse-information-retrieval
name: Sparse Information Retrieval
dataset:
name: NanoNQ
type: NanoNQ
metrics:
- type: dot_accuracy@1
value: 0.28
name: Dot Accuracy@1
- type: dot_accuracy@3
value: 0.6
name: Dot Accuracy@3
- type: dot_accuracy@5
value: 0.7
name: Dot Accuracy@5
- type: dot_accuracy@10
value: 0.78
name: Dot Accuracy@10
- type: dot_precision@1
value: 0.28
name: Dot Precision@1
- type: dot_precision@3
value: 0.2
name: Dot Precision@3
- type: dot_precision@5
value: 0.14
name: Dot Precision@5
- type: dot_precision@10
value: 0.078
name: Dot Precision@10
- type: dot_recall@1
value: 0.27
name: Dot Recall@1
- type: dot_recall@3
value: 0.58
name: Dot Recall@3
- type: dot_recall@5
value: 0.66
name: Dot Recall@5
- type: dot_recall@10
value: 0.72
name: Dot Recall@10
- type: dot_ndcg@10
value: 0.5073424422892974
name: Dot Ndcg@10
- type: dot_mrr@10
value: 0.4506666666666666
name: Dot Mrr@10
- type: dot_map@100
value: 0.4421389971626089
name: Dot Map@100
- task:
type: sparse-nano-beir
name: Sparse Nano BEIR
dataset:
name: NanoBEIR mean
type: NanoBEIR_mean
metrics:
- type: dot_accuracy@1
value: 0.3333333333333333
name: Dot Accuracy@1
- type: dot_accuracy@3
value: 0.5733333333333334
name: Dot Accuracy@3
- type: dot_accuracy@5
value: 0.6533333333333333
name: Dot Accuracy@5
- type: dot_accuracy@10
value: 0.7200000000000001
name: Dot Accuracy@10
- type: dot_precision@1
value: 0.3333333333333333
name: Dot Precision@1
- type: dot_precision@3
value: 0.25555555555555554
name: Dot Precision@3
- type: dot_precision@5
value: 0.19066666666666668
name: Dot Precision@5
- type: dot_precision@10
value: 0.13333333333333333
name: Dot Precision@10
- type: dot_recall@1
value: 0.197751243288938
name: Dot Recall@1
- type: dot_recall@3
value: 0.399004829640067
name: Dot Recall@3
- type: dot_recall@5
value: 0.476070114575135
name: Dot Recall@5
- type: dot_recall@10
value: 0.524806266613877
name: Dot Recall@10
- type: dot_ndcg@10
value: 0.44276319495624133
name: Dot Ndcg@10
- type: dot_mrr@10
value: 0.4641560846560846
name: Dot Mrr@10
- type: dot_map@100
value: 0.34058292349432834
name: Dot Map@100
---
# Inference-free SPLADE BERT-tiny trained on Natural-Questions tuples
This is a [Asymmetric Inference-free SPLADE Sparse Encoder](https://www.sbert.net/docs/sparse_encoder/usage/usage.html) model trained on the [natural-questions](https://huggingface.co/datasets/sentence-transformers/natural-questions) dataset using the [sentence-transformers](https://www.SBERT.net) library. It maps sentences & paragraphs to a 30522-dimensional sparse vector space and can be used for semantic search and sparse retrieval.
## Model Details
### Model Description
- **Model Type:** Asymmetric Inference-free SPLADE Sparse Encoder
<!-- - **Base model:** [Unknown](https://huggingface.co/unknown) -->
- **Maximum Sequence Length:** 512 tokens
- **Output Dimensionality:** 30522 dimensions
- **Similarity Function:** Dot Product
- **Training Dataset:**
- [natural-questions](https://huggingface.co/datasets/sentence-transformers/natural-questions)
- **Language:** en
- **License:** apache-2.0
### Model Sources
- **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
- **Documentation:** [Sparse Encoder Documentation](https://www.sbert.net/docs/sparse_encoder/usage/usage.html)
- **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
- **Hugging Face:** [Sparse Encoders on Hugging Face](https://huggingface.co/models?library=sentence-transformers&other=sparse-encoder)
### Full Model Architecture
```
SparseEncoder(
(0): Asym(
(query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast)
(corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM
(corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522})
)
)
```
## Usage
### Direct Usage (Sentence Transformers)
First install the Sentence Transformers library:
```bash
pip install -U sentence-transformers
```
Then you can load this model and run inference.
```python
from sentence_transformers import SparseEncoder
# Download from the 🤗 Hub
model = SparseEncoder("tomaarsen/inference-free-splade-bert-tiny-nq-3e-6-lambda-corpus")
# Run inference
sentences = [
'when did the american civil rights movement end',
'African-American civil rights movement (1954–1968) The Civil Rights Movement (also known as the American civil rights movement, African-American civil rights movement, and other terms,[b]) was a human rights movement from 1954–1968 that encompassed strategies, groups, and social movements to accomplish its goal of ending legalized racial segregation and discrimination laws in the United States. The movement secured the legal recognition and federal protection of black Americans in the United States Constitution and federal law.',
'Paleolithic Paleolithic humans made tools of stone, bone, and wood.[23] The early paleolithic hominins, Australopithecus, were the first users of stone tools. Excavations in Gona, Ethiopia have produced thousands of artifacts, and through radioisotopic dating and magnetostratigraphy, the sites can be firmly dated to 2.6Â\xa0million years ago. Evidence shows these early hominins intentionally selected raw materials with good flaking qualities and chose appropriate sized stones for their needs to produce sharp-edged tools for cutting.[29]',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# (3, 30522)
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
```
<!--
### Direct Usage (Transformers)
<details><summary>Click to see the direct usage in Transformers</summary>
</details>
-->
<!--
### Downstream Usage (Sentence Transformers)
You can finetune this model on your own dataset.
<details><summary>Click to expand</summary>
</details>
-->
<!--
### Out-of-Scope Use
*List how the model may foreseeably be misused and address what users ought not to do with the model.*
-->
## Evaluation
### Metrics
#### Sparse Information Retrieval
* Datasets: `NanoMSMARCO`, `NanoNFCorpus` and `NanoNQ`
* Evaluated with [<code>SparseInformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sparse_encoder/evaluation.html#sentence_transformers.sparse_encoder.evaluation.SparseInformationRetrievalEvaluator)
| Metric | NanoMSMARCO | NanoNFCorpus | NanoNQ |
|:-----------------|:------------|:-------------|:-----------|
| dot_accuracy@1 | 0.28 | 0.44 | 0.28 |
| dot_accuracy@3 | 0.54 | 0.58 | 0.6 |
| dot_accuracy@5 | 0.68 | 0.58 | 0.7 |
| dot_accuracy@10 | 0.74 | 0.64 | 0.78 |
| dot_precision@1 | 0.28 | 0.44 | 0.28 |
| dot_precision@3 | 0.18 | 0.3867 | 0.2 |
| dot_precision@5 | 0.136 | 0.296 | 0.14 |
| dot_precision@10 | 0.074 | 0.248 | 0.078 |
| dot_recall@1 | 0.28 | 0.0433 | 0.27 |
| dot_recall@3 | 0.54 | 0.077 | 0.58 |
| dot_recall@5 | 0.68 | 0.0882 | 0.66 |
| dot_recall@10 | 0.74 | 0.1144 | 0.72 |
| **dot_ndcg@10** | **0.5067** | **0.3143** | **0.5073** |
| dot_mrr@10 | 0.4317 | 0.5101 | 0.4507 |
| dot_map@100 | 0.4433 | 0.1363 | 0.4421 |
#### Sparse Nano BEIR
* Dataset: `NanoBEIR_mean`
* Evaluated with [<code>SparseNanoBEIREvaluator</code>](https://sbert.net/docs/package_reference/sparse_encoder/evaluation.html#sentence_transformers.sparse_encoder.evaluation.SparseNanoBEIREvaluator) with these parameters:
```json
{
"dataset_names": [
"msmarco",
"nfcorpus",
"nq"
]
}
```
| Metric | Value |
|:-----------------|:-----------|
| dot_accuracy@1 | 0.3333 |
| dot_accuracy@3 | 0.5733 |
| dot_accuracy@5 | 0.6533 |
| dot_accuracy@10 | 0.72 |
| dot_precision@1 | 0.3333 |
| dot_precision@3 | 0.2556 |
| dot_precision@5 | 0.1907 |
| dot_precision@10 | 0.1333 |
| dot_recall@1 | 0.1978 |
| dot_recall@3 | 0.399 |
| dot_recall@5 | 0.4761 |
| dot_recall@10 | 0.5248 |
| **dot_ndcg@10** | **0.4428** |
| dot_mrr@10 | 0.4642 |
| dot_map@100 | 0.3406 |
<!--
## Bias, Risks and Limitations
*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
-->
<!--
### Recommendations
*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
-->
## Training Details
### Training Dataset
#### natural-questions
* Dataset: [natural-questions](https://huggingface.co/datasets/sentence-transformers/natural-questions) at [f9e894e](https://huggingface.co/datasets/sentence-transformers/natural-questions/tree/f9e894e1081e206e577b4eaa9ee6de2b06ae6f17)
* Size: 99,000 training samples
* Columns: <code>query</code> and <code>corpus</code>
* Approximate statistics based on the first 1000 samples:
| | query | corpus |
|:--------|:-------------------|:-------------------|
| type | dict | dict |
| details | <ul><li></li></ul> | <ul><li></li></ul> |
* Samples:
| query | corpus |
|:---------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| <code>{'query': "who played the father in papa don't preach"}</code> | <code>{'corpus': 'Alex McArthur Alex McArthur (born March 6, 1957) is an American actor.'}</code> |
| <code>{'query': 'where was the location of the battle of hastings'}</code> | <code>{'corpus': 'Battle of Hastings The Battle of Hastings[a] was fought on 14 October 1066 between the Norman-French army of William, the Duke of Normandy, and an English army under the Anglo-Saxon King Harold Godwinson, beginning the Norman conquest of England. It took place approximately 7 miles (11 kilometres) northwest of Hastings, close to the present-day town of Battle, East Sussex, and was a decisive Norman victory.'}</code> |
| <code>{'query': 'how many puppies can a dog give birth to'}</code> | <code>{'corpus': 'Canine reproduction The largest litter size to date was set by a Neapolitan Mastiff in Manea, Cambridgeshire, UK on November 29, 2004; the litter was 24 puppies.[22]'}</code> |
* Loss: [<code>SpladeLoss</code>](https://sbert.net/docs/package_reference/sparse_encoder/losses.html#spladeloss) with these parameters:
```json
{'loss': SparseMultipleNegativesRankingLoss(
(model): SparseEncoder(
(0): Asym(
(query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast)
(corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM
(corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522})
)
)
(cross_entropy_loss): CrossEntropyLoss()
), 'lambda_corpus': 3e-06, 'lambda_query': 0, 'corpus_regularizer': FlopsLoss(
(model): SparseEncoder(
(0): Asym(
(query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast)
(corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM
(corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522})
)
)
), 'query_regularizer': None}
```
### Evaluation Dataset
#### natural-questions
* Dataset: [natural-questions](https://huggingface.co/datasets/sentence-transformers/natural-questions) at [f9e894e](https://huggingface.co/datasets/sentence-transformers/natural-questions/tree/f9e894e1081e206e577b4eaa9ee6de2b06ae6f17)
* Size: 1,000 evaluation samples
* Columns: <code>query</code> and <code>corpus</code>
* Approximate statistics based on the first 1000 samples:
| | query | corpus |
|:--------|:-------------------|:-------------------|
| type | dict | dict |
| details | <ul><li></li></ul> | <ul><li></li></ul> |
* Samples:
| query | corpus |
|:--------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| <code>{'query': 'where is the tiber river located in italy'}</code> | <code>{'corpus': 'Tiber The Tiber (/ˈtaɪbər/, Latin: Tiberis,[1] Italian: Tevere [ˈteːvere])[2] is the third-longest river in Italy, rising in the Apennine Mountains in Emilia-Romagna and flowing 406 kilometres (252\xa0mi) through Tuscany, Umbria and Lazio, where it is joined by the river Aniene, to the Tyrrhenian Sea, between Ostia and Fiumicino.[3] It drains a basin estimated at 17,375 square kilometres (6,709\xa0sq\xa0mi). The river has achieved lasting fame as the main watercourse of the city of Rome, founded on its eastern banks.'}</code> |
| <code>{'query': 'what kind of car does jay gatsby drive'}</code> | <code>{'corpus': 'Jay Gatsby At the Buchanan home, Jordan Baker, Nick, Jay, and the Buchanans decide to visit New York City. Tom borrows Gatsby\'s yellow Rolls Royce to drive up to the city. On the way to New York City, Tom makes a detour at a gas station in "the Valley of Ashes", a run-down part of Long Island. The owner, George Wilson, shares his concern that his wife, Myrtle, may be having an affair. This unnerves Tom, who has been having an affair with Myrtle, and he leaves in a hurry.'}</code> |
| <code>{'query': 'who sings if i can dream about you'}</code> | <code>{'corpus': 'I Can Dream About You "I Can Dream About You" is a song performed by American singer Dan Hartman on the soundtrack album of the film Streets of Fire. Released in 1984 as a single from the soundtrack, and included on Hartman\'s album I Can Dream About You, it reached number 6 on the Billboard Hot 100.[1]'}</code> |
* Loss: [<code>SpladeLoss</code>](https://sbert.net/docs/package_reference/sparse_encoder/losses.html#spladeloss) with these parameters:
```json
{'loss': SparseMultipleNegativesRankingLoss(
(model): SparseEncoder(
(0): Asym(
(query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast)
(corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM
(corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522})
)
)
(cross_entropy_loss): CrossEntropyLoss()
), 'lambda_corpus': 3e-06, 'lambda_query': 0, 'corpus_regularizer': FlopsLoss(
(model): SparseEncoder(
(0): Asym(
(query_0_IDF): IDF ({'frozen': False}, dim:30522, tokenizer: BertTokenizerFast)
(corpus_0_MLMTransformer): MLMTransformer({'max_seq_length': 512, 'do_lower_case': False}) with MLMTransformer model: BertForMaskedLM
(corpus_1_SpladePooling): SpladePooling({'pooling_strategy': 'max', 'activation_function': 'relu', 'word_embedding_dimension': 30522})
)
)
), 'query_regularizer': None}
```
### Training Hyperparameters
#### Non-Default Hyperparameters
- `eval_strategy`: steps
- `per_device_train_batch_size`: 64
- `per_device_eval_batch_size`: 64
- `learning_rate`: 2e-05
- `num_train_epochs`: 1
- `warmup_ratio`: 0.1
- `fp16`: True
- `batch_sampler`: no_duplicates
#### All Hyperparameters
<details><summary>Click to expand</summary>
- `overwrite_output_dir`: False
- `do_predict`: False
- `eval_strategy`: steps
- `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`: 2e-05
- `weight_decay`: 0.0
- `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`: 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
- `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`: no_duplicates
- `multi_dataset_batch_sampler`: proportional
</details>
### Training Logs
| Epoch | Step | Training Loss | Validation Loss | NanoMSMARCO_dot_ndcg@10 | NanoNFCorpus_dot_ndcg@10 | NanoNQ_dot_ndcg@10 | NanoBEIR_mean_dot_ndcg@10 |
|:------:|:----:|:-------------:|:---------------:|:-----------------------:|:------------------------:|:------------------:|:-------------------------:|
| 0.0129 | 20 | 1.5408 | - | - | - | - | - |
| 0.0259 | 40 | 1.4811 | - | - | - | - | - |
| 0.0388 | 60 | 1.2964 | - | - | - | - | - |
| 0.0517 | 80 | 0.9822 | - | - | - | - | - |
| 0.0646 | 100 | 0.6764 | - | - | - | - | - |
| 0.0776 | 120 | 0.547 | - | - | - | - | - |
| 0.0905 | 140 | 0.4755 | - | - | - | - | - |
| 0.1034 | 160 | 0.4212 | - | - | - | - | - |
| 0.1164 | 180 | 0.4562 | - | - | - | - | - |
| 0.1293 | 200 | 0.4057 | 0.3378 | 0.4848 | 0.3101 | 0.4742 | 0.4230 |
| 0.1422 | 220 | 0.3772 | - | - | - | - | - |
| 0.1551 | 240 | 0.3516 | - | - | - | - | - |
| 0.1681 | 260 | 0.3768 | - | - | - | - | - |
| 0.1810 | 280 | 0.3357 | - | - | - | - | - |
| 0.1939 | 300 | 0.3209 | - | - | - | - | - |
| 0.2069 | 320 | 0.3221 | - | - | - | - | - |
| 0.2198 | 340 | 0.3183 | - | - | - | - | - |
| 0.2327 | 360 | 0.3182 | - | - | - | - | - |
| 0.2456 | 380 | 0.333 | - | - | - | - | - |
| 0.2586 | 400 | 0.2946 | 0.2770 | 0.5115 | 0.3062 | 0.4842 | 0.4340 |
| 0.2715 | 420 | 0.295 | - | - | - | - | - |
| 0.2844 | 440 | 0.3019 | - | - | - | - | - |
| 0.2973 | 460 | 0.2882 | - | - | - | - | - |
| 0.3103 | 480 | 0.3203 | - | - | - | - | - |
| 0.3232 | 500 | 0.3215 | - | - | - | - | - |
| 0.3361 | 520 | 0.3018 | - | - | - | - | - |
| 0.3491 | 540 | 0.2918 | - | - | - | - | - |
| 0.3620 | 560 | 0.3365 | - | - | - | - | - |
| 0.3749 | 580 | 0.2847 | - | - | - | - | - |
| 0.3878 | 600 | 0.3382 | 0.2605 | 0.5192 | 0.3093 | 0.5002 | 0.4429 |
| 0.4008 | 620 | 0.2845 | - | - | - | - | - |
| 0.4137 | 640 | 0.2529 | - | - | - | - | - |
| 0.4266 | 660 | 0.2885 | - | - | - | - | - |
| 0.4396 | 680 | 0.2853 | - | - | - | - | - |
| 0.4525 | 700 | 0.2711 | - | - | - | - | - |
| 0.4654 | 720 | 0.2331 | - | - | - | - | - |
| 0.4783 | 740 | 0.2612 | - | - | - | - | - |
| 0.4913 | 760 | 0.2751 | - | - | - | - | - |
| 0.5042 | 780 | 0.2458 | - | - | - | - | - |
| 0.5171 | 800 | 0.2829 | 0.2475 | 0.5167 | 0.3117 | 0.5019 | 0.4434 |
| 0.5301 | 820 | 0.2698 | - | - | - | - | - |
| 0.5430 | 840 | 0.2455 | - | - | - | - | - |
| 0.5559 | 860 | 0.2769 | - | - | - | - | - |
| 0.5688 | 880 | 0.2569 | - | - | - | - | - |
| 0.5818 | 900 | 0.2404 | - | - | - | - | - |
| 0.5947 | 920 | 0.2538 | - | - | - | - | - |
| 0.6076 | 940 | 0.2449 | - | - | - | - | - |
| 0.6206 | 960 | 0.2649 | - | - | - | - | - |
| 0.6335 | 980 | 0.271 | - | - | - | - | - |
| 0.6464 | 1000 | 0.2081 | 0.2382 | 0.5087 | 0.3114 | 0.5082 | 0.4427 |
| 0.6593 | 1020 | 0.2627 | - | - | - | - | - |
| 0.6723 | 1040 | 0.2519 | - | - | - | - | - |
| 0.6852 | 1060 | 0.2463 | - | - | - | - | - |
| 0.6981 | 1080 | 0.2565 | - | - | - | - | - |
| 0.7111 | 1100 | 0.2586 | - | - | - | - | - |
| 0.7240 | 1120 | 0.2521 | - | - | - | - | - |
| 0.7369 | 1140 | 0.2441 | - | - | - | - | - |
| 0.7498 | 1160 | 0.2595 | - | - | - | - | - |
| 0.7628 | 1180 | 0.2612 | - | - | - | - | - |
| 0.7757 | 1200 | 0.2623 | 0.2324 | 0.5019 | 0.3129 | 0.5073 | 0.4407 |
| 0.7886 | 1220 | 0.2393 | - | - | - | - | - |
| 0.8016 | 1240 | 0.2606 | - | - | - | - | - |
| 0.8145 | 1260 | 0.2328 | - | - | - | - | - |
| 0.8274 | 1280 | 0.271 | - | - | - | - | - |
| 0.8403 | 1300 | 0.2556 | - | - | - | - | - |
| 0.8533 | 1320 | 0.2468 | - | - | - | - | - |
| 0.8662 | 1340 | 0.2389 | - | - | - | - | - |
| 0.8791 | 1360 | 0.2354 | - | - | - | - | - |
| 0.8920 | 1380 | 0.2331 | - | - | - | - | - |
| 0.9050 | 1400 | 0.2345 | 0.2303 | 0.5073 | 0.3139 | 0.5072 | 0.4428 |
| 0.9179 | 1420 | 0.2364 | - | - | - | - | - |
| 0.9308 | 1440 | 0.2125 | - | - | - | - | - |
| 0.9438 | 1460 | 0.2634 | - | - | - | - | - |
| 0.9567 | 1480 | 0.259 | - | - | - | - | - |
| 0.9696 | 1500 | 0.2496 | - | - | - | - | - |
| 0.9825 | 1520 | 0.2563 | - | - | - | - | - |
| 0.9955 | 1540 | 0.2475 | - | - | - | - | - |
| -1 | -1 | - | - | 0.5067 | 0.3143 | 0.5073 | 0.4428 |
### Environmental Impact
Carbon emissions were measured using [CodeCarbon](https://github.com/mlco2/codecarbon).
- **Energy Consumed**: 0.030 kWh
- **Carbon Emitted**: 0.012 kg of CO2
- **Hours Used**: 0.096 hours
### Training Hardware
- **On Cloud**: No
- **GPU Model**: 1 x NVIDIA GeForce RTX 3090
- **CPU Model**: 13th Gen Intel(R) Core(TM) i7-13700K
- **RAM Size**: 31.78 GB
### Framework Versions
- Python: 3.11.6
- Sentence Transformers: 4.2.0.dev0
- Transformers: 4.49.0
- PyTorch: 2.6.0+cu124
- Accelerate: 1.5.1
- Datasets: 2.21.0
- Tokenizers: 0.21.1
## 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",
}
```
#### SpladeLoss
```bibtex
@misc{formal2022distillationhardnegativesampling,
title={From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective},
author={Thibault Formal and Carlos Lassance and Benjamin Piwowarski and Stéphane Clinchant},
year={2022},
eprint={2205.04733},
archivePrefix={arXiv},
primaryClass={cs.IR},
url={https://arxiv.org/abs/2205.04733},
}
```
<!--
## Glossary
*Clearly define terms in order to be accessible across audiences.*
-->
<!--
## Model Card Authors
*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
-->
<!--
## Model Card Contact
*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
-->