all-MiniLM-L12-v2

This is a sentence-transformers model: It maps sentences & paragraphs to a 384 dimensional dense vector space and can be used for tasks like clustering or semantic search.

Evaluation Results

For an automated evaluation of this model, see the Sentence Embeddings Benchmark: https://seb.sbert.net

---

Background

The project aims to train sentence embedding models on very large sentence level datasets using a self-supervised contrastive learning objective. We used the pretrained microsoft/MiniLM-L12-H384-uncased model and fine-tuned in on a 1B sentence pairs dataset. We use a contrastive learning objective: given a sentence from the pair, the model should predict which out of a set of randomly sampled other sentences, was actually paired with it in our dataset.

We developped this model during the Community week using JAX/Flax for NLP & CV, organized by Hugging Face. We developped this model as part of the project: Train the Best Sentence Embedding Model Ever with 1B Training Pairs. We benefited from efficient hardware infrastructure to run the project: 7 TPUs v3-8, as well as intervention from Googles Flax, JAX, and Cloud team member about efficient deep learning frameworks.

Intended uses

Our model is intented to be used as a sentence and short paragraph encoder. Given an input text, it ouptuts a vector which captures the semantic information. The sentence vector may be used for information retrieval, clustering or sentence similarity tasks.

By default, input text longer than 256 word pieces is truncated.

Training procedure

Pre-training

We use the pretrained microsoft/MiniLM-L12-H384-uncased model. Please refer to the model card for more detailed information about the pre-training procedure.

Fine-tuning

We fine-tune the model using a contrastive objective. Formally, we compute the cosine similarity from each possible sentence pairs from the batch. We then apply the cross entropy loss by comparing with true pairs.

Hyper parameters

We trained ou model on a TPU v3-8. We train the model during 100k steps using a batch size of 1024 (128 per TPU core). We use a learning rate warm up of 500. The sequence length was limited to 128 tokens. We used the AdamW optimizer with a 2e-5 learning rate. The full training script is accessible in this current repository: train_script.py.

Training data

We use the concatenation from multiple datasets to fine-tune our model. The total number of sentence pairs is above 1 billion sentences. We sampled each dataset given a weighted probability which configuration is detailed in the data_config.json file.

Dataset	Paper	Number of training tuples
Reddit comments (2015-2018)	paper	726,484,430
S2ORC Citation pairs (Abstracts)	paper	116,288,806
WikiAnswers Duplicate question pairs	paper	77,427,422
PAQ (Question, Answer) pairs	paper	64,371,441
S2ORC Citation pairs (Titles)	paper	52,603,982
S2ORC (Title, Abstract)	paper	41,769,185
Stack Exchange (Title, Body) pairs	-	25,316,456
Stack Exchange (Title+Body, Answer) pairs	-	21,396,559
Stack Exchange (Title, Answer) pairs	-	21,396,559
MS MARCO triplets	paper	9,144,553
GOOAQ: Open Question Answering with Diverse Answer Types	paper	3,012,496
Yahoo Answers (Title, Answer)	paper	1,198,260
Code Search	-	1,151,414
COCO Image captions	paper	828,395
SPECTER citation triplets	paper	684,100
Yahoo Answers (Question, Answer)	paper	681,164
Yahoo Answers (Title, Question)	paper	659,896
SearchQA	paper	582,261
Eli5	paper	325,475
Flickr 30k	paper	317,695
Stack Exchange Duplicate questions (titles)		304,525
AllNLI (SNLI and MultiNLI	paper SNLI, paper MultiNLI	277,230
Stack Exchange Duplicate questions (bodies)		250,519
Stack Exchange Duplicate questions (titles+bodies)		250,460
Sentence Compression	paper	180,000
Wikihow	paper	128,542
Altlex	paper	112,696
Quora Question Triplets	-	103,663
Simple Wikipedia	paper	102,225
Natural Questions (NQ)	paper	100,231
SQuAD2.0	paper	87,599
TriviaQA	-	73,346
Total		1,170,060,424