TheRealAmoo 🤖🧪
@MehradAnsari
Followers
437
Following
2K
Media
40
Statuses
337
Teaching computers how to do science. ChE Ph.D. from @UofR. I try to advocate for open science and equality here.
Joined June 2020
RT @IYakavets: 🚨 Postdoc Opportunity! 🚨.Join us to develop an AI-powered self-driving lab for multiscale mapping of human tissue systems! 🧬….
0
5
0
Super excited to give a SPOTLIGHT talk on our chemist AI agent, #dZiner, at #NeurIPS2024 #AI4Mat on Sat@4:30pm. Feel free to stop by and hi. Preprint: Code:
1
0
8
Congrats to @BenBlaiszik for this leading this awesome event. Super grateful and excited to have been a small part of this global effort!.
For those of you looking for a big helping of LLMs for Thanksgiving, here you go! 🤖🦃 (link in next post). In this work, we release 34 examples (each with open code!) demonstrating applications of LLMs in materials science and chemistry across: .🔶molecular and material property.
0
0
7
At some point in life, I just wanna chill and write single-author papers like this 😅🍻.
0
0
5
dZiner's human-in-the-loop design can be simply done by chatting with the model on Slack , if you don't wanna worry about pip install. Here's an example of optimizing the binding affinity of a molecule against the DRD2 target. Details of the tools used are reported in the end.
Super excited to share our latest work at @acceleration_c!. dZiner: an agentic framework that can help with the closed-loop or human-in-the-loop language-to-structure materials inverse design. See thread 👇. Preprint: Code:
0
1
3
Just realized you can actually compile python scripts in latex. No fancy python library imports tho, cause there's apparently no pip install option. Would be great for automated table processing from local csv files. Maybe a future feature @overleaf ? 😃.
github.com
Compiling python code on latex. Contribute to mehradans92/py-latex development by creating an account on GitHub.
0
0
1
dZiner will appear in #NeurIPS-2024 #AI4Mat this year!🥳🥳🥳.Looking forward to your feedbacks and seeing everyone is Vancouver! 😃.
Super excited to share our latest work at @acceleration_c!. dZiner: an agentic framework that can help with the closed-loop or human-in-the-loop language-to-structure materials inverse design. See thread 👇. Preprint: Code:
2
1
4
dZiner can work with different textual representations or different surrogate models. See examples:.Like the GCNN CMC predictor by @AmySYQin and Victor Zavala: or the peptide hemolytic activity predictor by me and @andrewwhite01:.
1
0
0
Or a closed-loop optimization of MOF's organic linker for higher CO2 adsorption capacity. We fine-tuned MOFormer (by @ZhonglinJC and @AmirBaratiF) as the surrogate. You even can add extra design constraints in natural language text (Keep MW low, or don't use Fluoroamines, etc).
2
1
2
Here’s a human-AI collaboration example for optimizing a molecule for better binding against WDR5 protein target. Docking scores are estimated with AutoDock Vina (by @OlegTrott)
2
0
0
dZiner retrieves optimization guidelines for a target property from literature or other resources, suggests a new molecule, checks its chemical feasibility, and finally evaluates its property with uncertainty estimation using an ensemble of surrogate models.
1
0
0
Super excited to share our latest work at @acceleration_c!. dZiner: an agentic framework that can help with the closed-loop or human-in-the-loop language-to-structure materials inverse design. See thread 👇. Preprint: Code:
2
3
9
RT @steven_torrisi: Pictured here are four generations of interns brought together by @acceleration_c-- Linda Hung to my left, who mentored….
0
2
0
RT @felix_s_k: The "AI + Robotics" field has exploded in recent years. To stay on top of what's happening, we've put together a comprehensi….
0
8
0
RT @CoryMSimon: chillin’ at the Van art gallery with @steven_torrisi, @CoppLab, @MehradAnsari, and @ClaudianeOP! at the Accelerate Conferen….
0
1
0
Poster 3: Spock: LLM-based Tool for Bibliometric Analysis of Scientific Literature. By Youssef Briki, Siwoo Lee and Mehrad Ansari. Presented by Youssef Briki.
0
0
0
Poster 2: Active Learning Guided High-throughput Synthesis and Characterization of Mixed Metal Oxides. By Yutong Liu, Kangming Li, Robert Black, Mehrad Ansari, and Jason Hattrick-Simpers. Presented by Yutong Liu.
1
0
1
Poster 1: Flexible AI & Automation in Self-Driving Chemistry Labs. By Kourosh Darvish, Mehrad Ansari, and Sergio Pablo-García. Presented by the three of us at the AI & Automation Lab.
1
0
0