We're excited to introduce 🔥Yi-VL-34B🔥, newly open-sourced Yi Vision Language Models from . It now ranks
#1
in open-source category per MMMU & CMMMU benchmarks. Now our eyes 👀are wide open to see your fantastic vision projects!!
Silent for a while cos we've been busy making LLMs!! Coming up on🗓️May 13 will release a new suite of upgraded open-source models on
@huggingface
+ first proprietary model 𝗬𝗶-𝗟𝗮𝗿𝗴𝗲. Preview Yi-Large on
@lmsysorg
Chatbot Arena
New!🔥Yi-9B🔥has been open-sourced from
@01AI_Yi
. It stands out as the top-performing similar-sized language model friendly to developers, excelling in code and math. Welcome to give it a try and share how you solve problems!
Our team at
@01AI_Yi
is very proud to introduce the release of Yi-34B model now on top of
@huggingface
pretrained LLM leaderboard! Also a Yi-6B available. Welcome to give a try and build fantastic projects!
𝗬-𝗟𝗮𝗿𝗴𝗲 global debut from today. Experience our 🔥largest model🔥 in multiple ways:
Yi-Large API (global)
Yi-Large API (China)
Yi-Large + Wanzhi productivity product (万知 in China)
🚀Yi-34B-200K🚀has enhanced the long text capability. In the Needle-in-a-Haystack test, the performance rises from 89.3% to 99.8%.
@01AI_Yi
continues to pretrain the model on 5B tokens long-context data mixture and demonstrates a near-all-green performance!
team worked hard to launch Yi-34B-Chat finetuned on world's
#1
open source base model, 4 bits & 8 bits quantized versions also went 🔥live🔥 on
@huggingface
. More to build your LLM projects!
Yi models TECH REPORT just went live! Sharing our humble explorations launching, improving, innovating our base, chat, and vision-language models. Kudos to
@01AI_Yi
team behind the scenes. Love to hear the feedback from the community!
We are humbled by the evaluations from both
@huggingface
listing Yi-34B
#1
English/global open source LLM (pretrained) and C-Eval
#1
for Chinese open LLM. Quantized and chat model coming up soon. Share with more fellow developer friends!
Yi-Large-Preview ranks only 2 places under the newest
@OpenAI
@ChatGPTapp
GPT-4o on
@alpacaml
Eval2.0 verified category. Not bad, we'd say. Your thought? 😉 ➡️ Yi-Large API beta sign up:
“Yi: Open Foundation Models” by has arrived. Take a look into the specifics with us and discover the groundbreaking extensions of our Yi model series - which include long context, vision language, depth upscaling, and more. (1/6)
Wondering what's powering the trending
#yi34b
? Curious about the tech behind it? At
@01AI_Yi
, we've written a tech report revealing the inside scoop on:
✅Base and chat models
✅200K long context model
✅Depth-upscaled model
✅Vision-language model
Enjoy!
HuggingChat can now speak Chinese with Yi-34B-Chat e.g.🈷️ = "Moon" Yi open-source models on
@huggingface
are Chinese-English bilingual for you to build with the two most spoken languages in the world!! Try by changing current model setting to Yi on
Hugging Chat can now speak Chinese :-) Welcome to the first Chinese speaking model Yi-1.5-34B by
@01AI_Yi
on !
Let me know which assistant you built with this model :-D
My Markdown translator powered by Yi Feedback welcome!
Exciting results of Yi-34B-Chat fine-tuned model on AlpacaEval benchmark. With a solid base model, our AI Alignment team did a fine job. Still on the way to RL and more upgrades to share soon!
@huggingface
just updated the leaderboard with much harder evals. Yi-34B is now the top model of the leaderboard across all sizes/model categories!!! Thanks
@clefourrier
for sending the news. Would love to share Yi Open-source with more fellow developers
Ready for the *biggest update* of the Open LLM Leaderboard yet?
We just spent A YEAR of GPU time to make it more interesting and fairer! 🤯
How? With
@nathanhabib1011
, we added 3 new evals from the great EleutherAI harness 💥 and re-ran 2000+ models! 🚀
So, what changes? 🧵
A big shout out to the
@huggingface
community for helping us spot an oversight. We are in the process of publishing a new version. THANK YOU for helping out with our open-source debut in the past 1.5 weeks!!
Hi
@01AI_Yi
users!
🚀 Exciting announcement: Yi's APIs are now open for beta testing!
🔥 Ready to be a pioneer and unlock the power of Yi Models?
🙌 To gain free access, simply fill out this form.
⏳ Access is limited—Secure your spot now!
#Yi34B
#LLM
Kudos to Stanford teams for rolling out 👏👏AlpacaEval2.0!! Very impressed that Yi-34B-Chat by is closing the gap. Check out how you can build more projects from our open-source offers
Major upgrade: AlpacaEval->AlpacaEval2.0
✅ 7x faster
✅ 3x cheaper
✅ GPT-4 turbo baseline
✅ less length bias
✅ new models (Mixtral, Gemini,...)
We've annotated 12 models and hope to add 80. Seeking community support for annotations (~$400
@OpenAI
credits). Can anyone help?
Impressed by Stanford HELM new evals validating Yi-34B performance in accuracy. We encourage more developers to try and build interesting LLM projects! Check out Yi model open source offers on
Hello
#GTC2024
!! Impressive work done by amazing infra team, presented as a
@nvidia
official best practice for end-to-end FP8 training and inference. Our larger model is under way 💪💥
Yi Learning Hub is now live for the Yi community! You can find all the latest Yi model tech content at
Have you created tech content for Yi? If yes, we'd love to see more of your sharing by adding it to Yi Learning Hub. Prizes will go to top contributors!
Agree that this is THE essential chart on AI 2023. chose open-source as our debut, believing better AI can be built by more developers and used by more people. Yi models available in base, finetuned, and 4/8bits quantized versions
This is perhaps one of the most important charts on AI for 2024.
It was built by the amazing researcher team at
@CathieDWood
’s
@ARKInvest
.
We can see the rise of open source local models are on the path to overtake massive (and expensive) cloud based closed models.
Hi Yi users,
🌍 As
@01AI_Yi
is gaining traction, many of you are seeking a dedicated place for deeper conversations.
🔗 What's your preferred way to connect?
🗳️ Please vote & spread the word! You'll shape the channel!
📅 Don't miss out! Poll ends on Dec 22.
#Yi34B
#LLM
@erhartford
So glad the change was so very well received!
@erhartford
glad to have you on Yi open-source journey since day 1. Keep us posted if you make more builds with Yi-1.5!
🌟 Final Thoughts: (5/6)
In this report, we discuss the full-stack development of the Yi language model family. Key takeaways highlight the importance of data quantity and quality, along with our unique finetuning procedure outlined in the paper.
🙌We are shaping the
#YiUserGroup
identity and want to hear your voice.
🙋🏻♀️Which logo do you prefer?
💕Your opinion decides! Cast your vote and leave your mark on Yi's future now!
⏰Poll will be closed on Jan 27 at 23:59.
🔍 Long Context: (2/6)
The long-context capability offered by the Yi model series (Yi-34B-200K and Yi-6B-200K) has gained immense traction from the community. Learn how we extended the Yi base model to 200K long-context through various methods outlined in the paper.
🖼️ Visional Language: (3/6)
Earlier this year, we launched the Yi Vision Language models (Yi-VL-34B, Yi-VL-6B) with bilingual multimodal understanding and generation capabilities. Check out the architecture of the Yi-VL models and our three-stage training process.
@_philschmid
Thank you, we are glad to be here making a better product for the open source community, super excited to see what can build with Yi-1.5. Love to hear your feedback!
Thank you for your incredible enthusiasm about Yi's APIs!
🌊Our beta program has reached capacity, but for the lucky ones with access, get ready for an exhilarating adventure!
👀Don't forget to keep an eye out for more thrilling updates from Yi – there's plenty more to come!
Hi
@01AI_Yi
users!
🚀 Exciting announcement: Yi's APIs are now open for beta testing!
🔥 Ready to be a pioneer and unlock the power of Yi Models?
🙌 To gain free access, simply fill out this form.
⏳ Access is limited—Secure your spot now!
#Yi34B
#LLM
📈 Depth Upscaling: (4/6)
Scaling laws pose significant challenge in resource allocation. Discover our novel approach to dynamically adjust resource allocation, leading to the upscaling of Yi-6B base model to the Yi-9B base model with enhanced training efficiency & performance.
@markopolojarvi
During this beta testing phase:
1. The servers are located in Singapore.
2. Data sent to the Yi APIs will not be retained or used to train or improve Yi models.