fajie yuan

@sacdallago @KevinKaichuang x.com/i/status/20252…

0

6

738

Jason Hickel@jasonhickel·26 Şub

China's performance on Nature's university ranking is astonishing. After Harvard, every single university in the top 10 is Chinese. Wuhan University—in the city that Westerners considered an unknown outpost—outperforms Oxford, Cambridge, Yale, Cornell, Columbia and Chicago.

English

150

661

2.7K

158.3K

fajie yuan@duguyuan·27 Şub

Available Colab Links: ColabSaprot (35M,650M): colab.research.google.com/github/westlak… ColabSeprot (ESM1b‑650M, ESM2‑35M/150M/650M, ProTrek‑35M/650M, ProtBERT‑420M): colab.research.google.com/github/westlak… ColabESMC (300M, 600M): colab.research.google.com/github/westlak… ColabESM3 (1.4B): colab.research.google.com/github/westlak…

QME

125

fajie yuan@duguyuan·27 Şub

@sacdallago We have built ColabPLM with @KevinKaichuang and many other scientists providing various PLMs finetuning, including sequence models ESM1/2/C, ProtBert, ProtT5, ProTrek and structure model SaProt & ESM3. People can run them easily on the new data as baselines with zero coding.

English

0

1

212

Christian Dallago@sacdallago·27 Şub

Five years ago, we released FLIP. The core question was: can ML models for protein fitness prediction generalize in the ways that actually matter for protein engineering, i.e. low data, extrapolation to more mutations, out-of-distribution sequences?

English

3

10

56

4.6K

fajie yuan@duguyuan·23 Şub

@MusicalPlanet3 You can also share your model with one click after training on ColabSeprot or ColabSaprot. Then, all researchers can use your model or build new models based on it, and do not forget mention your work when sharing the model so that others can cite when using it.

English

Asutosh Rath@MusicalPlanet3

0

1

39

fajie yuan retweetledi

Asutosh Rath@MusicalPlanet3·23 Şub

@duguyuan This is super interesting work. Fine tuning and debugging ESM-650M took a lot of time as I was developing a tool for protein solubility last week. x.com/i/status/20258…

I built a protein solubility prediction tool which runs on ESM-650M, fine tuned on UESolDS dataset. Feel free to check it out. github.com/61-Keys/oracle…

English

182

fajie yuan@duguyuan·21 Şub

Want to fine-tune protein language models but don't have ML experience? 💻❌ We've got you covered! 📢 ✅ Previously: ColabSaprot & ColabSeprot (ESM1/2, ProTrek, ProtBert) 🆕 Now Available: ColabESMC & ColabESM3 Links➡️github.com/westlake-repl/… Tutorial➡️youtube.com/watch?v=nmLtjl…

YouTube

ColabSaprot & SaprotHub are now in @NatureBiotech! 🧬 A user-friendly, no-code platform for training, sharing, and collaborating on protein language models. We also provide ColabSeprot, integrating ESM1b, ESM2, ProTrek, and ProtBert for the community. nature.com/articles/s4158…

English

7

18

81

15.4K

fajie yuan@duguyuan·22 Şub

@omnicui Thanks Cui :)

English

0

87

CUI LUN🇫🇷🇨🇳@omnicui·22 Şub

Congratulations and great work!

Want to fine-tune protein language models but don't have ML experience? 💻❌ We've got you covered! 📢 ✅ Previously: ColabSaprot & ColabSeprot (ESM1/2, ProTrek, ProtBert) 🆕 Now Available: ColabESMC & ColabESM3 Links➡️github.com/westlake-repl/… Tutorial➡️youtube.com/watch?v=nmLtjl…

English

0

176

fajie yuan@duguyuan·21 Şub

@bots_and_bits After training on Colab, one can easily share their model on Hugging Face with just one click. Others can directly use these shared models on our Colab platform, or re-train them with their own data, and then share them back to the Hub. Everything can be done with a few clicks.

English

0

1

57

Benjamin Perry@bots_and_bits·21 Şub

@duguyuan Does this integrate with hugging face at all?

English

0

1

40

fajie yuan@duguyuan·21 Şub

One can also deploy them on your own GPU server: github.com/westlake-repl/…

English

1

197

fajie yuan@duguyuan·21 Şub

More members are here： #plm-members-of-opmc" target="_blank" rel="nofollow noopener">github.com/westlake-repl/…

English

3

307

fajie yuan@duguyuan·21 Şub

Available Colab Links: ColabSaprot (35M,650M): colab.research.google.com/github/westlak… ColabSeprot (ESM1b‑650M, ESM2‑35M/150M/650M, ProTrek‑35M/650M, ProtBERT‑420M): colab.research.google.com/github/westlak… ColabESMC (300M, 600M): colab.research.google.com/github/westlak… ColabESM3 (1.4B): colab.research.google.com/github/westlak…

HT

2

0

4

656

fajie yuan@duguyuan·21 Şub

Don't just train—share & collaborate! 🤝 With ColabPLM, easily fine-tune models on your specific datasets without coding. Then, use our One-Click Sharing to publish on PLMHub. Make your models instantly available to community - more visibility & citations for your research! 📈

English

1

186

fajie yuan retweetledi

Oogie@oggii_0·16 Şub

Seedance 2.0 is on another level. This entire game trailer was generated with AI in just minutes and the results are absolutely breathtaking. The visuals, the polish, the quality… it’s next-gen creativity in action.

English

124

175

1.9K

575.7K

fajie yuan retweetledi

Qwen@Alibaba_Qwen·16 Şub

🚀 Qwen3.5-397B-A17B is here: The first open-weight model in the Qwen3.5 series. 🖼️Native multimodal. Trained for real-world agents. ✨Powered by hybrid linear attention + sparse MoE and large-scale RL environment scaling. ⚡8.6x–19.0x decoding throughput vs Qwen3-Max 🌍201 languages & dialects 📜Apache2.0 licensed 🔗Dive in: GitHub: github.com/QwenLM/Qwen3.5 Chat: chat.qwen.ai API：modelstudio.console.alibabacloud.com/ap-southeast-1… Qwen Code: github.com/QwenLM/qwen-co… Hugging Face: huggingface.co/collections/Qw… ModelScope: modelscope.cn/collections/Qw… blog: qwen.ai/blog?id=qwen3.5

English

271

880

5.4K

1.3M

fajie yuan@duguyuan·7 Şub

Evolla surpasses general LLMs (GPT-4o, DeepSeek) in accuracy and shows competitive zero-shot performance against supervised specialists. Critically, it uncovers remote functional relationships missed by BLAST & Foldseek, boosting discovery in the low-homology zone.

English

2

181

fajie yuan@duguyuan·7 Şub

GitHub： github.com/westlake-repl/… HuggingFace 10B: huggingface.co/westlake-repl/… HuggingFace 80B: huggingface.co/westlake-repl/… paper: biorxiv.org/content/10.110…

English

0

1

196

fajie yuan@duguyuan·7 Şub

Our Evolla (chat-protein.com) preprint is updated! We used it to find: 🧬 Eukaryotic signature proteins in Asgard archaea. 🌊 A novel deep-sea PETase that degrades plastic. Surprise! HuggingFace downloads surpassed 90,000 this month! 🔥 PAPER: biorxiv.org/content/10.110…

We release our protein chatGPT, Evola! 🌟 chat-protein.com Evola comes in two versions: 10B & 80B. The 80B model has a 1.3B Saprot encoder & a 70B LLaMA3 decoder. Trained on 546 protein question-text pairs with an 150 billion word tokens! 💡🔬 biorxiv.org/content/10.110…

English