Nawaf Alampara
@Iam_Nawaf_
Followers
121
Following
220
Media
1
Statuses
56
PhD student @jablonkagroup . Previously @iitbombay. Accelerating scientific discovery using AI.
Mumbai, India
Joined August 2016
Less can be more for predicting properties with large language models 🤔. Checkout the new preprint
arxiv.org
Predicting properties from coordinate-category data -- sets of vectors paired with categorical information -- is fundamental to computational science. In materials science, this challenge...
Scale isn't all we need. @Iam_Nawaf_ shows with Santiago Miret that for the modeling of categorical-coordinate data with language model there is a cliff - what we term the Coordinate-Category Cliff. Language models fail to utilize coordinate data effectively.
0
1
5
RT @kmjablonka: I think that most ML benchmarks might be measuring the wrong things entirely. We've been building evaluation tools for ML….
0
8
0
RT @kmjablonka: A project we started a long time ago with @MichaelPieler, @pschwllr and many others is now on Arxiv: The ChemPile - a massi….
0
11
0
RT @kmjablonka: Title grew longer, the benchmark shrank: . ChemBench is now out today in @NatureChemistry. https:….
0
8
0
RT @jablonkagroup: LAMA Lab at ICLR in Singapore! Come check out our posters at the AI4MAT workshop on Monday #IC….
0
2
0
RT @jablonkagroup: We're excited to present our posters at the AI4Mat Workshop today at #ICLR25 #AI4Mat #Singapore
0
1
0
RT @jablonkagroup: 🚀ChemBench just leveled up! We’re thrilled to announce the latest release of ChemBench—now smarter and smoother! Dive in….
0
9
0
RT @kmjablonka: Many of our benchmarks underwent a large revision in the last weeks. We now host HuggingFace spaces for them and the datase….
0
5
0
RT @jablonkagroup: 🚀Our revised MaCBench paper is now on arxiv! Key updates!.🌟Robust reproducibility: 5x experiment….
0
6
0
RT @kmjablonka: Chemists often combine many different techniques to elucidate structures. @adrian_mirza_ has been building a system that….
0
23
0
RT @kmjablonka: I'm super proud of the team for the massive effort that went into this revision. It has been the most intense revision I….
0
6
0
RT @jablonkagroup: 🚨ChemBench Update🚨.We have been hard at work over the past few weeks improving our benchmark for evaluating LLMs in chem….
0
12
0
RT @kmjablonka: For benchmarks, too, less is more. We are deleting a lot of the next version of ChemBench. Deleting questions lets us….
0
4
0
RT @jablonkagroup: 1/8 📊 Materials science has a wealth of untapped data in papers, but lacks curated datasets for #ML and innovative mater….
0
13
0
Can LLMs model materials?.Checkout our latest work MatText, we explore some of the interesting and crucial questions in modelling materials with LLMs 🤓.
Do Language Models Need More Than Text & Scale for Materials Modeling?. A: Norm is more parameters + data = better performance across tasks. 📈 But materials science is governed by physical laws. This means scaling laws for language models can differ here!.
0
1
6
RT @jablonkagroup: This is the ChemBench core team (@adrian_mirza_,.@Iam_Nawaf_, @kunchapu22, @kmjablonka) trying to crash another leaderbo….
0
4
0
🚀 Excited to share our debut project from @jablonkagroup, ChemBench. Dive in for more insights: we've compared all the LLMs! #ChemBench #Teamwork #llm #ai4science #Chemistry.
Are large language models superhuman chemists?. We have been wondering about this, too and developed the first automated and comprehensive benchmark for evaluating large language models in chemistry and let LLMs compete with human chemists 👩🏼🔬. 📝
0
0
9
RT @jablonkagroup: 🎄 Wishing everyone a very happy Christmas from LAMA-Lab! 🎅🔬 . We've had an incredible time together last few days, feast….
0
2
0