diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
index b3416e5..8ee5aab 100644
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -15,6 +15,6 @@ jobs:
with:
python-version: 3.x
- name: Install dependencies
- run: pip install mkdocs-material pillow cairosvg
+ run: pip install mkdocs-material pillow cairosvg mkdocs-embed-external-markdown
- name: Deploy docs
run: mkdocs gh-deploy --force
diff --git a/docs/projects/indonesian-sentence-embeddings.md b/docs/projects/indonesian-sentence-embeddings.md
deleted file mode 100644
index a1e2d55..0000000
--- a/docs/projects/indonesian-sentence-embeddings.md
+++ /dev/null
@@ -1,283 +0,0 @@
-# Indonesian Sentence Embeddings
-
-Inspired by [Thai Sentence Vector Benchmark](https://github.com/mrpeerat/Thai-Sentence-Vector-Benchmark), we decided to embark on the journey of training Indonesian sentence embedding models!
-
-
- 
-
-
-## Evaluation
-
-### Semantic Textual Similarity
-
-We believe that a synthetic baseline is better than no baseline. Therefore, we followed approached done in the Thai Sentence Vector Benchmark project and translated the [STS-B](https://github.com/facebookresearch/SentEval) dev and test set to Indonesian via Google Translate API. This dataset will be used to evaluate our model's Spearman correlation score on the translated test set.
-
-> You can find the translated dataset on [🤗 HuggingFace Hub](https://huggingface.co/datasets/LazarusNLP/stsb_mt_id).
-
-### Retrieval
-
-To evaluate our models' capability to perform retrieval tasks, we evaluate them on Indonesian subsets of MIRACL and TyDiQA datasets. In both datasets, the model's ability to retrieve relevant documents given a query is tested. We employ R@1 (top-1 accuracy), MRR@10, and nDCG@10 metrics to measure our model's performance.
-
-### Classification
-
-For text classification, we will be doing emotion classification and sentiment analysis on the EmoT and SmSA subsets of [IndoNLU](https://huggingface.co/datasets/indonlp/indonlu), respectively. To do so, we will be doing the same approach as Thai Sentence Vector Benchmark and simply fit a Linear SVC on sentence representations of our texts with their corresponding labels. Thus, unlike conventional fine-tuning method where the backbone model is also updated, the Sentence Transformer stays frozen in our case; with only the classification head being trained.
-
-Further, we will evaluate our models using the official [MTEB](https://github.com/embeddings-benchmark/mteb.git) code that contains two Indonesian classification subtasks: `MassiveIntentClassification (id)` and `MassiveScenarioClassification (id)`.
-
-### Pair Classification
-
-We followed [MTEB](https://github.com/embeddings-benchmark/mteb.git)'s PairClassification evaluation procedure for pair classification. Specifically for zero-shot natural language inference tasks, all neutral pairs are dropped, while contradictions and entailments are re-mapped as `0`s and `1`s. The maximum average precision (AP) score is found by finding the best threshold value.
-
-We leverage the [IndoNLI](https://huggingface.co/datasets/indonli) dataset's two test subsets: `test_lay` and `test_expert`.
-
-## Methods
-
-### (Unsupervised) SimCSE
-
-We followed [SimCSE: Simple Contrastive Learning of Sentence Embeddings](https://arxiv.org/abs/2104.08821) and trained a sentence embedding model in an unsupervised fashion. Unsupervised SimCSE allows us to leverage an unsupervised corpus -- which are plenty -- and with different dropout masks in the encoder, contrastively learn sentence representations. This is parallel with the situation that there is a lack of supervised Indonesian sentence similarity datasets, hence SimCSE is a natural first move into this field. We used the [Sentence Transformer implementation](https://www.sbert.net/examples/unsupervised_learning/README.html#simcse) of [SimCSE](https://github.com/princeton-nlp/SimCSE).
-
-### ConGen
-
-Like SimCSE, [ConGen: Unsupervised Control and Generalization Distillation For Sentence Representation](https://github.com/KornWtp/ConGen) is another unsupervised technique to train a sentence embedding model. Since it is in-part a distillation method, ConGen relies on a teacher model which will then be distilled to a student model. The original paper proposes back-translation as the best data augmentation technique. However, due to the lack of resources, we implemented word deletion, which was found to be on-par with back-translation despite being trivial. We used the [official ConGen implementation](https://github.com/KornWtp/ConGen) which was written on top of the Sentence Transformers library.
-
-### SCT
-
-[SCT: An Efficient Self-Supervised Cross-View Training For Sentence Embedding](https://github.com/mrpeerat/SCT) is another unsupervised technique to train a sentence embedding model. It is very similar to ConGen in its knowledge distillation methodology, but also supports self-supervised training procedure without a teacher model. The original paper proposes back-translation as its data augmentation technique, but we implemented single-word deletion and found it to perform better than our backtranslated corpus. We used the [official SCT implementation](https://github.com/mrpeerat/SCT) which was written on top of the Sentence Transformers library.
-
-## Models
-
-| Model | #params | Base/Student Model | Teacher Model | Train Dataset | Supervised |
-| --------------------------------------------------------------------------------------------------------------------------- | :-----: | --------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------ | :--------: |
-| [SimCSE-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/simcse-indobert-lite-base) | 12M | [IndoBERT Lite Base](https://huggingface.co/indobenchmark/indobert-lite-base-p1) | N/A | [Wikipedia](https://huggingface.co/datasets/LazarusNLP/wikipedia_id_20230520) | |
-| [SimCSE-IndoRoBERTa Base](https://huggingface.co/LazarusNLP/simcse-indoroberta-base) | 125M | [IndoRoBERTa Base](https://huggingface.co/flax-community/indonesian-roberta-base) | N/A | [Wikipedia](https://huggingface.co/datasets/LazarusNLP/wikipedia_id_20230520) | |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 125M | [IndoBERT Base](https://huggingface.co/indobenchmark/indobert-base-p1) | N/A | [Wikipedia](https://huggingface.co/datasets/LazarusNLP/wikipedia_id_20230520) | |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 12M | [IndoBERT Lite Base](https://huggingface.co/indobenchmark/indobert-lite-base-p1) | [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | [Wikipedia](https://huggingface.co/datasets/LazarusNLP/wikipedia_id_20230520) | |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 125M | [IndoBERT Base](https://huggingface.co/indobenchmark/indobert-base-p1) | [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | [Wikipedia](https://huggingface.co/datasets/LazarusNLP/wikipedia_id_20230520) | |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 125M | [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | [Wikipedia](https://huggingface.co/datasets/LazarusNLP/wikipedia_id_20230520) | |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 125M | [IndoBERT Base](https://huggingface.co/indobenchmark/indobert-base-p1) | [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | [Wikipedia](https://huggingface.co/datasets/LazarusNLP/wikipedia_id_20230520) | |
-| [S-IndoBERT Base mMARCO](https://huggingface.co/LazarusNLP/s-indobert-base-mmarco) | 125M | [IndoBERT Base](https://huggingface.co/indobenchmark/indobert-base-p1) | N/A | [mMARCO](https://huggingface.co/datasets/unicamp-dl/mmarco) | ✅ |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 134M | [DistilBERT Base Multilingual](https://huggingface.co/distilbert-base-multilingual-cased) | mUSE | See: [SBERT](https://www.sbert.net/docs/pretrained_models.html#model-overview) | ✅ |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | 125M | [XLM-RoBERTa Base](https://huggingface.co/xlm-roberta-base) | [paraphrase-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-mpnet-base-v2) | See: [SBERT](https://www.sbert.net/docs/pretrained_models.html#model-overview) | ✅ |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 118M | [Multilingual-MiniLM-L12-H384](https://huggingface.co/microsoft/Multilingual-MiniLM-L12-H384) | See: [arXiv](https://arxiv.org/abs/2212.03533) | See: [🤗](https://huggingface.co/intfloat/multilingual-e5-small) | ✅ |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 278M | [XLM-RoBERTa Base](https://huggingface.co/xlm-roberta-base) | See: [arXiv](https://arxiv.org/abs/2212.03533) | See: [🤗](https://huggingface.co/intfloat/multilingual-e5-base) | ✅ |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | 560M | [XLM-RoBERTa Large](https://huggingface.co/xlm-roberta-large) | See: [arXiv](https://arxiv.org/abs/2212.03533) | See: [🤗](https://huggingface.co/intfloat/multilingual-e5-large) | ✅ |
-
-## Results
-
-### Semantic Textual Similarity
-
-#### Machine Translated Indonesian STS-B
-
-| Model | Spearman's Correlation (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :--------------------------: |
-| [SimCSE-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/simcse-indobert-lite-base) | 44.08 |
-| [SimCSE-IndoRoBERTa Base](https://huggingface.co/LazarusNLP/simcse-indoroberta-base) | 61.26 |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 70.13 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 79.97 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 80.47 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 81.16 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 74.56 |
-| [S-IndoBERT Base mMARCO](https://huggingface.co/LazarusNLP/s-indobert-base-mmarco) | 72.95 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 75.08 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | **83.83** |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 78.89 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 79.72 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | 79.44 |
-
-### Retrieval
-
-#### MIRACL
-
-| Model | R@1 (%) ↑ | MRR@10 (%) ↑ | nDCG@10 (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :-------: | :----------: | :-----------: |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 36.04 | 48.25 | 39.70 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 46.04 | 59.06 | 51.01 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 45.93 | 58.58 | 49.95 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 45.83 | 58.27 | 49.91 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 40.41 | 47.29 | 40.68 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 41.35 | 54.93 | 48.79 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | 52.81 | 65.07 | 57.97 |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 68.33 | 78.85 | 73.84 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 68.95 | 78.92 | 74.58 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | **69.89** | **80.09** | **75.64** |
-
-#### TyDiQA
-
-| Model | R@1 (%) ↑ | MRR@10 (%) ↑ | nDCG@10 (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :-------: | :----------: | :-----------: |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 61.94 | 69.89 | 73.52 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 75.22 | 81.55 | 84.13 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 73.09 | 80.32 | 83.29 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 72.38 | 79.37 | 82.51 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 76.81 | 83.16 | 85.87 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 70.44 | 77.94 | 81.56 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | 81.41 | 87.05 | 89.44 |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 90.97 | 94.14 | 95.25 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 91.85 | 94.88 | 95.82 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | **94.15** | **96.36** | **97.14** |
-
-### Classification
-
-#### MTEB - Massive Intent Classification `(id)`
-
-| Model | Accuracy (%) ↑ | F1 Macro (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :------------: | :------------: |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 59.71 | 57.70 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 62.41 | 60.94 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 61.14 | 60.02 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 60.93 | 59.50 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 55.66 | 54.48 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 55.99 | 52.44 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | 65.43 | 63.55 |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 64.16 | 61.33 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 66.63 | 63.88 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | **70.04** | **67.66** |
-
-#### MTEB - Massive Scenario Classification `(id)`
-
-| Model | Accuracy (%) ↑ | F1 Macro (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :------------: | :------------: |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 66.14 | 65.56 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 67.25 | 66.53 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 67.72 | 67.32 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 67.12 | 66.64 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 61.89 | 60.97 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 65.25 | 63.45 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | 70.72 | 70.58 |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 67.92 | 67.23 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 70.70 | 70.26 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | **74.11** | **73.82** |
-
-#### IndoNLU - Emotion Classification (EmoT)
-
-| Model | Accuracy (%) ↑ | F1 Macro (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :------------: | :------------: |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 55.45 | 55.78 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 58.18 | 58.84 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 57.04 | 57.06 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 59.54 | 60.37 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 61.13 | 61.70 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 63.63 | 64.13 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | 63.18 | 63.78 |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 64.54 | 65.04 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 68.63 | 69.07 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | **74.77** | **74.66** |
-
-#### IndoNLU - Sentiment Analysis (SmSA)
-
-| Model | Accuracy (%) ↑ | F1 Macro (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :------------: | :------------: |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 85.6 | 81.50 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 81.2 | 75.59 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 85.4 | 82.12 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 83.0 | 78.74 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 82.0 | 76.92 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 78.8 | 73.64 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | 89.6 | **86.56** |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 83.6 | 79.51 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 89.4 | 86.22 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | **90.0** | 86.50 |
-
-### Pair Classification
-
-#### IndoNLI
-
-| Model | `test_lay` AP (%) ↑ | `test_expert` AP (%) ↑ |
-| --------------------------------------------------------------------------------------------------------------------------- | :-----------------: | :--------------------: |
-| [SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/simcse-indobert-base) | 56.06 | 50.72 |
-| [ConGen-IndoBERT Lite Base](https://huggingface.co/LazarusNLP/congen-indobert-lite-base) | 69.44 | 53.74 |
-| [ConGen-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-indobert-base) | 71.14 | 56.35 |
-| [ConGen-SimCSE-IndoBERT Base](https://huggingface.co/LazarusNLP/congen-simcse-indobert-base) | 70.80 | 56.59 |
-| [SCT-IndoBERT Base](https://huggingface.co/LazarusNLP/sct-indobert-base) | 59.82 | 53.41 |
-| [distiluse-base-multilingual-cased-v2](https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2) | 58.48 | 50.50 |
-| [paraphrase-multilingual-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2) | **74.87** | **57.96** |
-| [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 63.97 | 51.85 |
-| [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 60.25 | 50.91 |
-| [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | 61.39 | 51.62 |
-
-## References
-
-```bibtex
-@misc{Thai-Sentence-Vector-Benchmark-2022,
- author = {Limkonchotiwat, Peerat},
- title = {Thai-Sentence-Vector-Benchmark},
- year = {2022},
- publisher = {GitHub},
- journal = {GitHub repository},
- howpublished = {\url{https://github.com/mrpeerat/Thai-Sentence-Vector-Benchmark}}
-}
-```
-
-```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",
-}
-```
-
-```bibtex
-@inproceedings{gao2021simcse,
- title={{SimCSE}: Simple Contrastive Learning of Sentence Embeddings},
- author={Gao, Tianyu and Yao, Xingcheng and Chen, Danqi},
- booktitle={Empirical Methods in Natural Language Processing (EMNLP)},
- year={2021}
-}
-```
-
-```bibtex
-@inproceedings{limkonchotiwat-etal-2022-congen,
- title = "{ConGen}: Unsupervised Control and Generalization Distillation For Sentence Representation",
- author = "Limkonchotiwat, Peerat and
- Ponwitayarat, Wuttikorn and
- Lowphansirikul, Lalita and
- Udomcharoenchaikit, Can and
- Chuangsuwanich, Ekapol and
- Nutanong, Sarana",
- booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
- year = "2022",
- publisher = "Association for Computational Linguistics",
-}
-```
-
-```bibtex
-@article{10.1162/tacl_a_00620,
- author = {Limkonchotiwat, Peerat and Ponwitayarat, Wuttikorn and Lowphansirikul, Lalita and Udomcharoenchaikit, Can and Chuangsuwanich, Ekapol and Nutanong, Sarana},
- title = "{An Efficient Self-Supervised Cross-View Training For Sentence Embedding}",
- journal = {Transactions of the Association for Computational Linguistics},
- volume = {11},
- pages = {1572-1587},
- year = {2023},
- month = {12},
- issn = {2307-387X},
- doi = {10.1162/tacl_a_00620},
- url = {https://doi.org/10.1162/tacl\_a\_00620},
- eprint = {https://direct.mit.edu/tacl/article-pdf/doi/10.1162/tacl\_a\_00620/2196817/tacl\_a\_00620.pdf},
-}
-```
-
-## Credits
-
-Indonesian Sentence Embeddings is developed with love by:
-
-
\ No newline at end of file
diff --git a/docs/projects/machine-translation.md b/docs/projects/machine-translation.md
index 56b55b4..41044d3 100644
--- a/docs/projects/machine-translation.md
+++ b/docs/projects/machine-translation.md
@@ -1,105 +1,4 @@
-# Machine Translation
+- [:material-github: GitHub Repository](https://github.com/LazarusNLP/machine-translation)
+- [🤗 HuggingFace Collection](https://huggingface.co/collections/LazarusNLP/indot5-6541fbdfa385933e811c2e1f)
-## Indo-mT5
-
-Indo-mT5 is mT5 fine-tuned for machine translation of regional languages of Indonesia. We release our dataset creation scripts, training code, and fine-tuned models for other to leverage.
-
-There are two types of models:
-
-- **Multilingual**: Many-to-many, multilingual translation model.
-- **Bilingual**: Unidirectional, bilingual translation model.
-
-We also further experiment with two settings:
-
-- **Baseline**: Model trained on 7 languages (`ace`, `ban`, `bug`, `ind`, `jav`, `min`, `sun`).
-- **All**: Model trained on 45 languages as listed [here](languages.md).
-
-## Training
-
-Our experiments are conducted in these steps:
-
-- **Multilingual Training on Bible**: We first fine-tuned mT5 on multilingual translation on parallel Bible dataset, creating Indo-mT5.
-- **Multilingual Training on NusaX**: We take Indo-mT5 and fine-tune them on multilingual pairs of the NusaX dataset.
-- **Bilingual Training on NusaX**: We take Indo-mT5 and fine-tune them on bilingual pairs of the NusaX dataset.
-
-Therefore, we have six training scripts:
-
-| Dataset | Config | Type | Training Script | Evaluation Script |
-| ------- | -------- | ------------ | ---------------------------------------------------------------------------- | -------------------------------------------------------------------------- |
-| Bible | Baseline | Multilingual | [train_bible_baseline.sh](train_bible_baseline.sh) | [eval_bible_baseline.sh](eval_bible_baseline.sh) |
-| Bible | All (v2) | Multilingual | [train_bible_all.sh](train_bible_all.sh) | [eval_bible_all.sh](eval_bible_all.sh) |
-| NusaX | Baseline | Multilingual | [train_nusax_baseline_multilingual.sh](train_nusax_baseline_multilingual.sh) | [eval_nusax_baseline_multilingual.sh](eval_nusax_baseline_multilingual.sh) |
-| NusaX | All (v2) | Multilingual | [train_nusax_all_multilingual.sh](train_nusax_all_multilingual.sh) | [eval_nusax_all_multilingual.sh](eval_nusax_all_multilingual.sh) |
-| NusaX | Baseline | Bilingual | [train_nusax_baseline_bilingual.sh](train_nusax_baseline_bilingual.sh) | [eval_nusax_baseline_bilingual.sh](eval_nusax_baseline_bilingual.sh) |
-| NusaX | All (v2) | Bilingual | [train_nusax_all_bilingual.sh](train_nusax_all_bilingual.sh) | [eval_nusax_all_bilingual.sh](eval_nusax_all_bilingual.sh) |
-
-## Results
-
-We evaluated our models on NusaX (Winata et al., 2022) and compared them to existing models.
-
-### `ind -> x`
-
-| Model | #params | `ace` | `ban` | `bbc` | `bjn` | `bug` | `jav` | `mad` | `min` | `nij` | `sun` | avg |
-| ------------------------------------ | ------- | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: |
-| IndoGPT (Winata et al., 2022) | 117M | 9.60 | 14.17 | 8.20 | 22.23 | 5.18 | 24.05 | 14.44 | 26.95 | 17.56 | 23.15 | 16.55 |
-| IndoBART v2 (Winata et al., 2022) | 132M | **19.21** | **27.08** | **18.41** | **40.03** | **11.06** | **39.97** | 28.95 | 48.48 | **27.11** | **38.46** | 29.88 |
-| mBART-50 Large (Winata et al., 2022) | 610M | 17.21 | 22.67 | 17.79 | 34.26 | 10.78 | 35.33 | 28.63 | 43.87 | 25.91 | 31.21 | 26.77 |
-| mT5 Base (Winata et al., 2022) | 580M | 14.79 | 18.07 | 18.22 | 38.64 | 6.68 | 33.48 | 0.96 | 45.84 | 13.59 | 33.79 | 22.41 |
-| NLLB-200 Distilled (zero-shot) | 600M | 2.74 | 4.87 | - | - | 1.66 | 17.66 | - | 9.79 | - | 11.92 | 8.11 |
-| Indo-mT5 NusaX Multilingual | 580M | 16.02 | 22.48 | - | - | 8.86 | 33.65 | - | 33.65 | - | 29.76 | 24.07 |
-| Indo-mT5 NusaX Bilingual | 580M | 17.99 | 27.03 | - | - | 10.80 | 39.63 | - | **51.56** | - | 35.16 | **30.36** |
-| Indo-mT5 v2 NusaX Multilingual | 580M | 14.28 | 19.19 | 14.86 | 28.39 | 8.05 | 28.70 | 20.95 | 32.70 | 22.30 | 26.19 | 21.56 |
-| Indo-mT5 v2 NusaX Bilingual | 580M | 17.58 | 24.24 | 16.69 | 38.81 | 10.20 | 37.87 | **29.77** | 50.90 | 26.93 | 34.22 | 28.72 |
-
-### `x -> ind`
-
-| Model | #params | `ace` | `ban` | `bbc` | `bjn` | `bug` | `jav` | `mad` | `min` | `nij` | `sun` | avg |
-| ------------------------------------ | ------- | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: | :-------: |
-| IndoGPT (Winata et al., 2022) | 117M | 7.01 | 13.23 | 5.27 | 19.53 | 1.98 | 27.31 | 13.75 | 23.03 | 10.83 | 23.18 | 14.51 |
-| IndoBART v2 (Winata et al., 2022) | 132M | 24.44 | 40.49 | 19.94 | **47.81** | 12.64 | **50.64** | 36.10 | 58.38 | **33.50** | **45.96** | 36.99 |
-| mBART-50 Large (Winata et al., 2022) | 610M | 18.45 | 34.23 | 17.43 | 41.73 | 10.87 | 39.66 | 32.11 | 59.66 | 29.84 | 35.19 | 31.92 |
-| mT5 Base (Winata et al., 2022) | 580M | 18.59 | 21.73 | 12.85 | 42.29 | 2.64 | 45.22 | 32.35 | 58.65 | 25.61 | 36.58 | 29.65 |
-| NLLB-200 Distilled (zero-shot) | 600M | 9.42 | 21.24 | - | - | 6.18 | 30.54 | - | 40.49 | - | 26.91 | 22.46 |
-| Indo-mT5 NusaX Multilingual | 580M | 23.94 | 35.30 | - | - | **16.68** | 29.76 | - | 48.10 | - | 36.54 | 31.72 |
-| Indo-mT5 NusaX Bilingual | 580M | **24.78** | **42.15** | - | - | 16.27 | 47.26 | - | **62.94** | - | 42.39 | **39.30** |
-| Indo-mT5 v2 NusaX Multilingual | 580M | 21.01 | 30.43 | 18.57 | 34.21 | 14.42 | 35.19 | 27.04 | 42.64 | 26.90 | 33.78 | 28.42 |
-| Indo-mT5 v2 NusaX Bilingual | 580M | 22.87 | 39.48 | **20.48** | 44.53 | 15.97 | 45.20 | **36.65** | 60.97 | 32.38 | 39.80 | 35.83 |
-
-## Parallel Bible Dataset Creation
-
-This will cover the creation process of our Bible machine-translation dataset.
-
-### Overview
-
-1. Scrape Bible Data
-2. Align Bible Verses
-3. Load as Machine-Translation Dataset
-
-### Bible Scraping
-
-```sh
-python utils/scrape_parallel.py \
- --codes abun aceh ambdr aralle balantak bali bambam bauzi berik bugis dairi duri ende galela gorontalo iban jawa kaili_daa karo kupang lampung madura makasar mamasa manggarai mentawai meyah minang mongondow napu ngaju nias rote sabu sangir sasak simalungun sunda taa tabaru tb toba toraja uma yali yawa \
- --outdir corpus \
- -j 4
-```
-
-### Align Bible Verses
-
-```sh
-for LANGUAGE in abun aceh ambdr aralle balantak bali bambam bauzi berik bugis dairi duri ende galela gorontalo iban jawa kaili_daa karo kupang lampung madura makasar mamasa manggarai mentawai meyah minang mongondow napu ngaju nias rote sabu sangir sasak simalungun sunda taa tabaru tb toba toraja uma yali yawa
-do
- python utils/align.py --path corpus/$LANGUAGE.json --outdir corpus_aligned
-done
-```
-
-You can read more about aligning Bible verses in [our blogpost](https://lazarusnlp.github.io/blogs/bible_alignment/).
-
-### Data Loading Script
-
-In the data loading script, we have to do these two steps:
-
-1. Split unique verse IDs into train/test/validation subsets.
-2. Generate permutations of every verse ID for every subset.
-
-You can find our data loading implementation in [src/alkitab-sabda-mt.py](src/alkitab-sabda-mt.py).
\ No newline at end of file
+{{ external_markdown('https://raw.githubusercontent.com/LazarusNLP/machine-translation/main/README.md', '') }}
\ No newline at end of file
diff --git a/docs/projects/sentence-embeddings.md b/docs/projects/sentence-embeddings.md
new file mode 100644
index 0000000..36310fb
--- /dev/null
+++ b/docs/projects/sentence-embeddings.md
@@ -0,0 +1,5 @@
+- [:material-github: GitHub Repository](https://github.com/LazarusNLP/indonesian-sentence-embeddings)
+- [:material-web: Documentation](https://lazarusnlp.github.io/indonesian-sentence-embeddings/)
+- [🤗 HuggingFace Collection](https://huggingface.co/collections/LazarusNLP/indonesian-sentence-embedding-6541fce662e82d932ff360c5)
+
+{{ external_markdown('https://raw.githubusercontent.com/LazarusNLP/indonesian-sentence-embeddings/main/README.md', '') }}
\ No newline at end of file
diff --git a/mkdocs.yml b/mkdocs.yml
index b39dd00..7e331ce 100644
--- a/mkdocs.yml
+++ b/mkdocs.yml
@@ -24,9 +24,22 @@ theme:
- navigation.tabs
- navigation.tabs.sticky
# - navigation.sections
+ - navigation.top
+ # - toc.integrate
+
+nav:
+ - Home: index.md
+ - Blogs:
+ - Bible Alignment: blogs/bible_alignment.md
+ - Indonesian Accents and Regional Languages: blogs/accents_and_languages.md
+ - Projects:
+ - Sentence Embeddings: projects/sentence-embeddings.md
+ - Machine Translation: projects/machine-translation.md
plugins:
- social
+ - search
+ - external-markdown
markdown_extensions:
- attr_list
@@ -47,4 +60,5 @@ markdown_extensions:
pygments_lang_class: true
- pymdownx.inlinehilite
- pymdownx.snippets
+ - pymdownx.details
- pymdownx.superfences
-
-
-
-