Project MIRACL

Official handle for MIRACL: Multilingual Information Retrieval across a Continuum of Languages. For questions --- [email protected]

[Project MIRACL @project_miracl]

🌍🙌🌏 MIRACL is here!

The corpora + training/dev splits for all 16 known languages are now on HuggingFace!

🔥53k queries, with 470k human-labeled relevance judgments!

💥Find the data here: https://t.co/B5iNhW6Xlv

👉Link to paper: https://t.co/aavuJiXQvY

[Project MIRACL @project_miracl]

How was MIRACL made? Check out the methods from our winners🔥

NOT CIIR: https://t.co/xGyVmvkYwX
Bott: https://t.co/GZ99nNcnUn
NLE: https://t.co/apw8n66hQP

[Project MIRACL @project_miracl]

💥WSDM Cup Day tomorrow!
Location: Level 4, Esplanade 2

(UTC+8)
8:30-10:00 Keynote and Overview
13:30-15:00 Talks from our winners!
NOT CIIR:🥈in Known track @huang_zhiqi @pxyumass @jallan_umass
Bott:🥈in Surprise track🥉in Known track: Qi Zhang
NLE:🥇in BOTH tracks @cadurosar

[Project MIRACL @project_miracl]

📢📢 MIRACL Competition is now over!

We would like to thank all the participants. We received submissions from over 40 different teams participating for #WSDM2023 Cup! 🎉🎉

Our final leaderboard results can be viewed publicly below👇
https://t.co/yhCIuTt4Mr

[Jimmy Lin @lintool]

Recently, @CohereAI boasted "3X better performance" in multilingual text understanding. We tested that claim by evaluating Cohere embeddings on MIRACL:

tl;dr - We weren't able to replicate the 3X claim, but we did observe a 38% improvement over BM25. https://t.co/hfYSdL5C22

[Project MIRACL @project_miracl]

📢📢📢 Extra 4️⃣ days for submission!
MIRACL competition deadline is now extended to Jan. 16th (AoE)
Time to submit your models and results on the MIRACL Leaderboard! ⏰
Participate and win those amazing cash prizes in the two tasks! 🎁
Submit here👇
https://t.co/ENYHwcBpSL

[Project MIRACL @project_miracl]

💥Surprise languages and Test-B topics released💥
How good your model perform on German and Yoruba in MIRACL?

👉 Data: https://t.co/8vVoVzjjSq
👉 Baselines (for surprise language dev set): https://t.co/YFeUOu6s2H
(Reproducible on Pyserini)

🗓 One week left to win prizes🥇🥈🥉!

[Jimmy Lin @lintool]

Now that the semester is over and those GPUs will be idle during the holidays... what better time to work on MIRACL? #wsdm2023 cup challenge on search across 18 different languages https://t.co/fQtMDDCxxP Leaderboard: https://t.co/hfxBnNUyr4

[Project MIRACL @project_miracl]

🔥Topics of Test-A released!

💫 Available on HuggingFace Datasets

👉 https://t.co/8vVoVziM2S

⚡️Submit to leaderboard to view the Test-A scores!

[Project MIRACL @project_miracl]

🔥MIRACL 🌍🙌🌏 leaderboard is now up!

Submissions are welcome 🙌!

👉 https://t.co/ENYHwcBpSL

[RAGTIME Track 🔎 TREC 2026 @trec_ragtime]

If you enjoyed #NeuCLIR2022, you might be interested in @project_miracl, a shared task at #WSDM2023 on multilingual retrieval! Check it out:

[Project MIRACL @project_miracl]

🔥🔥🔥 A MIRACL is coming tomorrow!

🌍🙌🌏 77k queries, 700k judgments on Wikipedia in 16 languages!
🆕 train/dev sets drop tomorrow

Join our mailing list and Slack to stay up to date!
📬 Mailing List: https://t.co/oaDzp4CG1w
💬 Slack: https://t.co/UXAuSnqWH2

[Nandan Thakur @beirmug]

Our mailing list is out now! 🆕

If you're interested in participating in Project MIRACL 🌍🙌🌏 - come join our mailing list! 📬

For more details, checkout below 👇

[Jimmy Lin @lintool]

If you're interested in MIRACL 🌍🙌🌏 - come join our mailing list! https://t.co/sdmKbu5TS1

[Project MIRACL @project_miracl]

If you're interested in participating in Project MIRACL 🌍🙌🌏 - come join our mailing list! 📬

https://t.co/oaDzp4CG1w

- Prizes worth over 6k USD to be won!
- Opportunity to try out the largest multi-monolingual retrieval dataset!

[Project MIRACL @project_miracl]

Introducing MIRACL 🌍🙌🌏(Multilingual Information Retrieval Across a Continuum of Languages), an WSDM 2023 Cup challenge that focuses on search across 18 languages, which collectively encompass over three billion native speakers around the world. https://t.co/IexzhUNyYQ