This is an embedding model based on stella_en_1.5B_v5 and further fine-tuned for retrieval tasks in Polish. It transforms texts into 1024-dimensional vectors. The model training consisted of two stages:

In the first stage, we adapted the model to support the Polish language using the multilingual knowledge distillation method, leveraging a diverse corpus of 20 million Polish-English text pairs.
The original Stella model and the output of the first stage were limited to a short context of 512 tokens. In the second stage, we extended the context to 8192 tokens and then fine-tuned the model using contrastive loss on a dataset comprising 1.5 million queries. Positive and negative passages for each query have been selected with the help of BAAI/bge-reranker-v2.5-gemma2-lightweight reranker. The model was trained for three epochs with a batch size of 1024 queries.

Usage (Sentence-Transformers)

The model utilizes the same prompts as the original stella_en_1.5B_v5.

For retrieval, queries should be prefixed with "Instruct: Given a web search query, retrieve relevant passages that answer the query.\nQuery: ".

For symmetric tasks such as semantic similarity, both texts should be prefixed with "Instruct: Retrieve semantically similar text.\nQuery: ".

Please note that the model uses a custom implementation, so you should add trust_remote_code=True argument when loading it. You can use the model like this with sentence-transformers:

python

from sentence_transformers import SentenceTransformer
from sentence_transformers.util import cos_sim

model = SentenceTransformer(
    "sdadas/stella-pl-retrieval-8k",
    trust_remote_code=True,
    device="cuda"
)
model.bfloat16()

# Retrieval example
query_prefix = "Instruct: Given a web search query, retrieve relevant passages that answer the query.\nQuery: "
queries = [query_prefix + "Jak dożyć 100 lat?"]
answers = [
    "Trzeba zdrowo się odżywiać i uprawiać sport.",
    "Trzeba pić alkohol, imprezować i jeździć szybkimi autami.",
    "Gdy trwała kampania politycy zapewniali, że rozprawią się z zakazem niedzielnego handlu."
]
queries_emb = model.encode(queries, convert_to_tensor=True, show_progress_bar=False)
answers_emb = model.encode(answers, convert_to_tensor=True, show_progress_bar=False)
best_answer = cos_sim(queries_emb, answers_emb).argmax().item()
print(answers[best_answer])

# Semantic similarity example
sim_prefix = "Instruct: Retrieve semantically similar text.\nQuery: "
sentences = [
    sim_prefix + "Trzeba zdrowo się odżywiać i uprawiać sport.",
    sim_prefix + "Warto jest prowadzić zdrowy tryb życia, uwzględniający aktywność fizyczną i dietę.",
    sim_prefix + "One should eat healthy and engage in sports.",
    sim_prefix + "Zakupy potwierdzasz PINem, który bezpiecznie ustalisz podczas aktywacji."
]
emb = model.encode(sentences, convert_to_tensor=True, show_progress_bar=False)
print(cos_sim(emb, emb))

Evaluation Results

The model achieves NDCG@10 of 62.69 on the Polish Information Retrieval Benchmark. See PIRB Leaderboard for detailed results.

Acknowledgements

The research was supported [in part] by project “Cloud Artificial Intelligence Service Engineering (CAISE) platform to create universal and smart services for various application areas”, No. KPOD.05.10-IW.10-0005/24, as part of the European IPCEI-CIS program, financed by NRRP (National Recovery and Resilience Plan) funds. Computations were carried out using the computers of Centre of Informatics Tricity Academic Supercomputer & Network at Gdansk University of Technology.

Citation

bibtex

@inproceedings{dadas2024pirb,
  title={PIRB: A Comprehensive Benchmark of Polish Dense and Hybrid Text Retrieval Methods},
  author={Dadas, Slawomir and Pere{\l}kiewicz, Micha{\l} and Po{\'s}wiata, Rafa{\l}},
  booktitle={Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)},
  pages={12761--12774},
  year={2024}
}