• mle
@cohere
working on llm posttraining🫡 • prev:
@UCR_CSE
&
@BilkentUniv
• i sing & play bass & write stuff • tweets in 🇹🇷🇬🇧🎸 • flowers in my hair 🌸🌺
for context, i have a lakland skyline bass rn.
in the video, my collaborator Art Bronzie and i are just jamming and having good fun. 🫡🫡
listen to our collabs on Spotify:
This is an incredible video for everyone who would like to internalize how Fast Fourier Transform works. FFT is at the core of numerous technological advancements and is also a brilliant algorithm.
Personal update: I started working at
@CohereAI
as a machine learning engineer today. I am so grateful for the opportunity and excited for what lies ahead 🤖🎉🥳✨
Basvurmak icin son birkac gun! ML kariyerinizi ucuracak bu firsati kacirmayin derim. 🚀🚀🚀🚀
Ben Cohere'de calisiyorum; sorularinizi bana iletebilirsiniz ✨
One of the most exciting aspects of the full-time paid Scholars Program is that we will accept remote candidates from around the world.
We look forward to collaborating with early-career ML researchers across LatAm, Europe, and Africa! 4 days to apply.
We are excited to launch the next cohort of the Cohere For AI Scholars Program — an opportunity to work with some of the best researchers in the world. 🎉 The Scholars Program is designed to change where, how, and by whom research is done. Apply by Aug 30!
Join us to help release a new MMLU test set that better represents global models. 🌍 We’re seeking researchers to help us verify translations in 39 target languages.
⚒️Task:
Just finished typing up a new research proposal. I will announce it tomorrow at
@forai_ml
🌞✨ Sooooo excited for this 🙈🎉🥳 I love being a part of this community sm 🫶🏼🥺
Announcement: my book Deep Learning with Python (2nd edition) has been released.
500 pages of code examples, theory, context, practical tips... If you want to really understand how deep learning works, why it matters, and how to use it, this is your book!
@apieceofcosmos
buna karsi cikicam mlsf. matematik sadece fizige degil, butun bilim ve muhendislik dallarina hizmet eden bir alandir. mesela matematik olmasa bu tiviti (guvenli bir sekilde) gonderemezdik. kriptografinin fizikte pek kullanildigini sanmiyorum ve tamamen matematik temelli bir alan
Turkiye'den adaylarin rahatlikla basvurabilecegi, ML/AI alaninda konusunda oldukca degerli tecrubeler kazandirdigini dusundugum bu programa ilgilenen herkesin basvurmasini tavsiye ederim. Sorulariniz olursa yazabilirsiniz✨
Not: Calisan link thread'in sonunda 🙏
We are excited to launch the next cohort of the Cohere For AI Scholars Program — an opportunity to work with some of the best researchers in the world. 🎉 The Scholars Program is designed to change where, how, and by whom research is done. Apply by Aug 30!
Selamlar, ben
@cohere
'de ML munendisi olarak calisiyorum. Dogal dil modelleri gelistiriyoruz. Kendimizi gun be gun iyilestirdigimiz, tatmin edici bir calisma tempomuz var. Ise alimlarimiz mevcut, basvurmayi dusunun derim:
⌘R+
Welcoming Command R+, our latest model focused on scalability, RAG, and Tool Use. Like last time, we're releasing the weights for research use, we hope they're useful to everyone!
🚨 🚨 ⌘R+
Our new state-of-the-art LLM. It is multilingual, amazing at RAG and tool-use 🔥
Enjoy its open weights as well. Super proud to be part of
@cohere
@CohereForAI
.
Today is the day we share AYA with the world! Aya is a multilingual open-source LLM finetuned on over 513M instruction-style data points, which we also release as open-source! Aya is a state-of-the-art multilingual model that supports 101 languages and performs better than
📣📣 Announcing a new RAG-focused multilingual LLM with open weights! We have been working on it for a while at
@cohere
and are excited to share that it is finally out!!!!
Announcing C4AI Command R+ open weights, a state-of-the-art 104B LLM with RAG, tooling and multilingual in 10 languages.
This release builds on our 35B and is a part of our commitment to make AI breakthroughs accessible to the research community. 🎉
Congratulations to head of Cohere For AI and VP of Research at
@cohere
,
@sarahookr
, for being recognized as part of
@TIME
's list TIME100 AI of 2024!
#time100ai
Today, we introduce Coral: a knowledge assistant for enterprises looking to improve the productivity of their most strategic teams. Users can converse with Coral to help them complete their business tasks.
Today, we’re launching Aya, a new open-source, massively multilingual LLM & dataset to help support under-represented languages. Aya outperforms existing open-source models and covers 101 different languages – more than double covered by previous models.
🏆 ACL Best Paper Award:
- Causal Estimation of Memorisation Profiles by Lesci et al.
- Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model by Üstün et al.
#NLProc
#ACL2024NLP
Haftaya Pazartesi
@irombie
ile NLP şirketlerinde çalışmak, MLE mülakat süreçleri ve genel olarak ML Mühendisliği üzerine konuşacağız. Siz de merak ettiklerinizi sormak ve İrem’in bu alandaki deneyimlerinden yararlanmak isterseniz bize katılın:
#NLP
#MLE
Ah, you are too kind 😌😌
but yes, I'll be in Istanbul on July 13, giving a workshop on ML/AI (details TBA but I can assure you it will be fun and interesting 🤓)
sign up here 👉👉
⌘-R
Introducing Command-R, a model focused on scalability, RAG, and Tool Use. We've also released the weights for research use, we hope they're useful to the community!
Aya Model paper wins a Best Paper Award at ACL, 2024.
We're honoured that our initiative to push the boundaries of multilingual AI and scientific practices has been so prestigiously recognized.
Thank you to everyone who has believed in Aya. 🌍
life has been incredibly tough on me for the last 3 months. there were times i thought i would not make it, but i somehow powered through. i experienced first hand that human beings are incredibly resilient.
hoping that 2024 will bring nothing but goodness for all of us.🙏
We’re excited to announce our Series D financing to accelerate growth, expand our team, & develop our next class of frontier, enterprise-grade, data privacy-focused AI technology.
We’re bringing highly scalable and secure AI solutions to global enterprises wherever their data is
I started a new blog, called 2cute2tech (), which I hope will be a safe and fun space for everyone who is interested to write blog posts on their technical expertise. I am also more than happy to collaborate on posts, so reach out!
@alarasclimax
duyar kasmak istemiyorum fakat shein oldukca kotu bir marka. arastirip okuduklarima gore hem calisanlarini havasiz ortamlarda uzun saatler calistiriyor, hem bunca kalitesiz urunu hizlica piyasaya salarak cevre kirliligine neden oluyor, hem de indie tasarimcilarin tasarimlarini
We’re celebrating our first year of Cohere For AI and the moments that made this year special! Here’s a look back at the highlights at Cohere For AI on our first anniversary! 🧵
@CohereForAI
'in AYA projesi var. 101 farkli dilden soru cevap formatinda veri toplayip bunu acik kaynak olarak yayinlayacaklar. Bu verileri gonulluler giriyor. Turkce geride maalesef. Guzel bir grup olusturup haftada bir saat bulusup hem sosyallessek hem de veri girsek aslinda👇
We released several works last week which I am very proud of. 🎉
C4AI is a small lab and these research directions take a lot of time + care. Sharing here a little of the thinking behind each one.
If you want to learn more about our work:
Today,
@cohere
's non-profit research lab,
@CohereForAI
, has released their open-source multilingual LLM, Aya.
It has been trained on 101 languages and built with 3,000 collaborators across 119 countries with Instruction Fine-Tuning (IFT).
I've tried it in Turkish.
Aya is
Buyuk chatbot modellerinin nasil gelistirildigi ve Cohere API'ini kullanarak bir Discord bot'u yapmak ile ilgili iki konusma yapacagim bu etkinligi dort gozle bekliyorum 🤩🤩 dogal dil isleme ve chatbotlar ilginizi cekiyorsa kaydolun derim! 🌞
Kadın Yazılımcı olarak uzun süredir üzerinde çalıştığımız Diva: Dive into AI Konferansı’nı,
@TeknasyonTech
ana sponsorluğunda 19 Ağustos Cumartesi günü İstanbul Bahçeşehir Üniversitesi Beşiktaş Kampüsü’nde gerçekleştireceğiz! 🎊
Size heyecanli bir haberle daha geldim! Eger projeye duzenli bir sekilde katki saglarsaniz bu buyuk acik kaynak projeye katki sagladiginiza dair sertifika alacaksiniz. Bana cokca soru soruluyor "CV'mde ne guzel durur?" diye. Bu sertifika CV'nizde harika gorunecektir!
@CohereForAI
'in AYA projesi var. 101 farkli dilden soru cevap formatinda veri toplayip bunu acik kaynak olarak yayinlayacaklar. Bu verileri gonulluler giriyor. Turkce geride maalesef. Guzel bir grup olusturup haftada bir saat bulusup hem sosyallessek hem de veri girsek aslinda👇
Our work called "Sparsified Secure Aggregation for Privacy-Preserving Federated Learning" is available on arxiv. In this thread, I will talk about the concepts in this paper and explain what we do in this work. (1/n)
Wrote a huuuuuge chunk of my paper, fixed a serious bug that drastically improved my experimental results, and almost finished the first draft of the paper even though there are more than 3 weeks left until the deadline. On fire today 💃💃💃💃💃
Katilan herkese cok tesekkur ederim, sizlerle birlikte olmak cok keyifliydi. Ileride baska etkinliklerde gorusmek uzere! organizasyon icin
@kadinyazilimci
'ya da cok tesekkurler
@abilgegunduz
ben ne zaman hayatimda stresli bir donem gecirsem kafeini birakiyorum. vucutta kortizol fazlasi varken kafein baya rahatsiz ediyo, kaygilari arttiriyo asiri. kahveyi assiri seven biri olarak o donemlerde decaf takiliorum
📢DJANGO BİLEN YAZILIMCI ARANIYOR: Ankara tabipler odası şu anda deprem bölgelerindeki sağlık çalışanlarının ve hastanelerin koordine olabilmesi için merkezi bi sistem oluşturuyor. Sıkıntılar şunlar: 👇🏻(1/4)
We’re celebrating our second year of Cohere For AI with 🔟 ways in which we have delighted in exploring the unknown. 🎉Here’s a look back on C4AI highlights on our second-year anniversary! ✨
Glad to share that I was accepted bu the finance track of OxML summer school as an in person participant! Super excited to learn more about ML x Finance! 🥳🥳
The
@aclmeeting2024
conference in Bangkok, Thailand is less than a week away and we are excited to share what our Cohere and C4AI teams will be presenting.🙌
Check out where we are throughout the conference here:
Meet the 2021 Delta teaching fellows!
@crypt_irem
is a Ph.D. candidate at UC Riverside. Her research interests include privacy preserving machine learning, federated learning, and machine learning for social good.
Subscribe for her upcoming talk 📺⬇️
this is so true, i can't even 😮💨
evals are often reduced to a set of benchmarks designed for tasks significantly different than how humans normally use large language models, such as solving multiple choice questions. we should not focus on hillclimbing on metrics so out of
Honestly surprised we still have to repeat this message. h/t
@FanaHOVA
Repeat after me:
I will build evals for my tasks.
I will build evals for my tasks.
I will build evals for me tasks.
Today, we’re excited to release Command-R, a new RAG-optimized LLM aimed at large-scale production workloads.
Command-R fits into the emerging “scalable” category of models that balance high efficiency with strong accuracy, enabling companies to move beyond proof of concept, and
Our work called "Secure Aggregation for Privacy-Aware Federated Learning with Limited Resources" has been accepted at ICLR 2022 Workshop on Socially Responsible Machine Learning 🌟🙌🏼🥳
Two years ago today I joined
@cohere
to bring back as
@CohereForAI
💙
Thank you to everyone who has supported our work in the two years since. I am a bit in awe of everything that has happened since.
Cheers to exploring the unknown, together. 🍾✨
One of the best part of this job is meeting students from the past 😊. İrem Ergün (
@irombie
) is an ML engineer at
@CohereForAI
. She was kind enough to come and visit my undergraduate ML course
@sabanciu
. She even helped us design a contemporary homework on LLMs.
I will start a technical blog soon. The first blog post will be about federated learning since it is one my research areas. I plan to publish the same content in both English and Turkish. Done with more than half of the blog post and looking forward to making it online!
Proud to be selected as a Global Teaching Fellow at
@deltanalytics
🥳 We just had our kickoff meeting, and I met wonderful people!
Thank you for the opportunity. Thrilled to be a part of this community, having such a powerful mission to make ML more accesible on a global scale!
dogal dil isleme konusunda arastirma yapmak isteyen ve henuz cok arastirma tecrubesi olmayan insanlara dolu dolu arastirma yaptiracak program basvurularini bekliyor! inanilmaz yetenekli arastirmacilarla birlikte calisacaginiz muhtesem bir firsat, kacirmayin derim! 🎉🎉
Today we are thrilled to launch the second cohort of our Cohere For AI Scholars Program — an opportunity to work with some of the best researchers in the world. This Program is designed to change where, how, and by whom research is done. Apply by Sept. 11!
Today, we’re excited to announce new funding to help us bring generative AI to enterprises around the globe.
AI will be the heart that powers the next decade of business success, and Cohere is ready to lead the way.
I have exciting news! I finally finished the first federated learning tutorial. 🥳
I will make it a sequence, so we will cover quite a bit! Please give me feedback if you think some things should be changed so we can learn and improve together!
@forai_ml
acik kaynak cok dilli bir chatbot gelistirmek icin bir insiyatif baslatti. Adi Aya. Turkce hedefledikleri dillerden biri. Bu modelde Turkcenin de temsil edilmesine katki saglamak isterseniz diye bunu paylasmak istedim:
Hadi yarisalim 😈
I have been getting a lot of questions about synthethic data these days. Highly recommend reading this paper if you are interested in synthetic data generation ✨
Arbitrage is the future.
Out of all the strategies we have tried for accelerating multilingual performance -- this has led to dramatic 56% relative gains in win-rates.
Very proud we are releasing this work -- which will spur massive gains for Aya 23.
📜:
There’s less than 1 week until
@iclr_conf
Conference in Vienna, Austria 🥳
We’ll be showcasing our research, presenting at workshops and connecting with our community. Here’s where you can find the C4AI &
@Cohere
team throughout the conference this week.
Participated in my first programming competition today!🤓🤓 I did not do well at all.😅However, I see this as an opportunity to learn and grow, and I plan to attend the weekly competitive programming meetings @ UCR to level up my programming game.
🎉 Milestone alert: the C4AI Community has hit the 1000+ members mark! 🎉
These 1000+ ML researchers hail from 95 countries and engage in discussions, attend meetings & events, and collaborate on research.
We're honoured to learn, grow, and explore with you all. Here's to YOU!
I will be presenting my poster called "Distributed Privacy-Preserving Machine Learning" today between 1pm-2:30pm PDT at North America School of Information Theory! Super excited!!
Irem -
@irombie
dün akşam Public Speaking Mentorship programımızın konuğuydu. Katılımcıların çok faydalandığı bir sunum yaptı. Kaçıranlar ve tekrar izlemek isteyenler için sunumunu yayınladık:
Today we are thrilled to launch the second cohort of our Cohere For AI Scholars Program — an opportunity to work with some of the best researchers in the world. This Program is designed to change where, how, and by whom research is done. Apply by Sept. 11!
📣 ACIK KAYNAK BIR PROJE ICIN TURKCE SORU CEVAP FORMATINDA VERI OLUSTURMAYA HEVESLI GONULLULER ARANIYOR
⚠️ ONEMLI LINKLER:
✅ Aya:
👾 Discord sunucusu katilim linki:
Detaylar 👇🏻👇🏻👇🏻👇🏻👇🏻👇🏻
@CohereForAI
'in Aya isimli bir
Aya - Closing the Contribution Chapter.
Join us to mark this moment in AI history. 🌌
On Dec 15, 9am ET we're hosting a live stream with stories of resilience and celebration from some of our 3,000 collaborators across 117 countries.
🗓️
We're pleased to share that Aya 23 - 8B is now hosted and available via the Cohere API. 🔥
Our hope is that this will continue to make multilingual generative AI breakthroughs more accessible to the research community. 🌍
We are hiring a machine learning engineer role to drive making our research + weight releases as accessible as possible to the wider community. 🔥
If you care about model efficiency, tooling, usability, translating research into impact -- get in touch!