Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura) header
Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura) profile picture

PhD @Harvard | AI & Robotics ⚡ Health | Developer of OpenBioLLM, Open Medical-LLM Leaderboard, MedMCQA, Med-Halt | I like high-quality datasets & eval 🧠

Cambridge, MA
Joined December 2021
Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)Photo by Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ (@aadityaura)
Introducing OpenBioLLM-Llama3-70B & 8B: The most capable openly available Medical-domain LLMs to date! 🩺💊🧬 Outperforms industry giants like GPT-4, Gemini, Meditron-70B, Med-PaLM-1, and Med-PaLM-2 in the biomedical domain. 🏥📈 🌟 OpenBioLLM-70B delivers SOTA performance, setting a new state-of-the-art for models of its size. OpenBioLLM-8B model and even surpasses GPT-3.5, Gemini, and Meditron-70B! 🚀 Today's release is just the beginning! In the coming months, we'll be introducing: - Expanded medical domain coverage 🧠 - Longer context windows 📜🔍 - Better benchmarks 📈🏆 - Multimodal capabilities🖥️🩺📊🔬 Medical-LLM Leaderboard: huggingface.co/spaces/openli… #gpt #gpt4 #gemini #medical #llm #chatgpt #opensource #llama3 #meta
30
197
21
816
162,074
This is super cool!! 🔥 Been following Tanishq's work for a while, and really happy to see @iScienceLuvr and the Sophont team pushing the boundaries of what's possible with pathology AI. So much potential here for researchers. Huge congrats on the launch, Tanishq! 🚀
Cancer patients with the same diagnosis often have very different tumor biology. Molecular testing can reveal underlying differences, but that's slow and expensive. Introducing Holon-S1, our platform to predict genes and biomarkers from routine pathology slides. • beats two other models at predicting spatial gene activity in a controlled comparison • has identified potentially novel biomarkers in breast cancer and colon cancer • evaluated across 16 organ types and more than 15 scanners, using slides from both formalin-fixed, paraffin-embedded and fresh-frozen tissue. We believe Holon-S1 has the potential to help researchers find new biomarkers in slides they already have, discover treatment targets, and bring more personalized cancer care within reach!
1
10
934
Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ retweeted
HealthBench Professional, our new evaluation of clinician chat tasks, is now available on HuggingFace for easy access. Each example was written, reviewed, and adjudicated by three or more physicians. We sampled heavily for difficulty for recent OpenAI models, so the evaluation should remain relevant for upcoming models. huggingface.co/datasets/open…
12
29
4
211
28,919
Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ retweeted
We built an AI system that discovers health biomarkers from wearable data. One of its first findings: "late-night doomscrolling" is a statistically validated predictor of depression severity (ρ = 0.177, p < 0.001, n = 7,497). The AI named the feature. No human guidance.
21
100
15
633
80,245
Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ retweeted
The attention mechanism in Transformers just got a major upgrade 🧠 📄 arxiv.org/abs/2602.21371 Introducing Interleaved Head Attention - instead of keeping your attention heads independent, share information between attention heads for improved reasoning and compositionality. Pseudo-queries, pseudo-keys, and pseudo-values as learned as linear combinations of the original heads, so interaction across heads happens before attention is computed. IHA stays compatible with efficient implementations like FlashAttention ⚡ Significant improvements on reasoning and long-context tasks: 📊 +5.8% on GSM8K (Maj@16) 📊 +2.8% on MATH-500 (Maj@16) 📊 +27% at 4k on RULER 📊 +32% at 8k 📊 +112% at 16k Theoretical results show that IHA strictly generalizes multi-head attention.
🧵 [1/14]: Talking-Heads Attention by @NoamShazeer et al. showed something interesting: maybe attention heads shouldn’t be fully isolated. 🧠 That got us thinking: If communication across heads matters, what is the right way for heads to communicate, especially from a one-layer reasoning perspective? 🔗⚙️ That question led us to Interleaved Head Attention (IHA) ✨ 📄 Paper link: arxiv.org/pdf/2602.21371
7
46
1
368
52,403
Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ retweeted
We started our journey as a precision robotics company with a clear mission – advancing industrial automation to the microscale. This video highlights one of the key requirements for making that mission possible: repeatability below 1 micrometer. #robotics #robotarm #engineering
10
51
7
418
30,412
Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ retweeted
Touch was the missing sense in robotics But FlexiTac is here to change that. An open-source, low-cost, scalable tactile sensor that’s: • ~3 min to fabricate • ~$2.5 per unit • Real-time + ML-ready This is what making hardware actually accessible looks like.
3
24
5
128
13,050
Aaditya Ura @ NeurIPS 2026 (Atlanta) ✈️ retweeted
LoRA makes fine-tuning more accessible, but it's unclear how it compares to full fine-tuning. We find that the performance often matches closely---more often than you might expect. In our latest Connectionism post, we share our experimental results and recommendations for LoRA. thinkingmachines.ai/blog/lor…
83
553
177
3,467
1,460,332
🚨 🔥 New Paper Alert: Presenting ReXVQA - One of the largest & most comprehensive benchmarks for Visual Question Answering in Radiology 🚨 - 694K+ questions - 160K chest X-ray - Evaluated across 8 state-of-the-art multimodal LLMs Thread 🧵 #MedicalAI #Radiology #MedGemma
1
3
12
1,020
This paper has been accepted at PSB 2026! 🔥🚀 🙌A big shoutout to collaborator & reviewers @ Jung oh Lee, @XiaomanZhang99 @malai_san @pranavrajpurkar for insightful and constructive feedback on this work. Preprint: arxiv.org/abs/2506.04353 Dataset: huggingface.co/datasets/rajp…
1
1
5
173
✨ Big life update: Happy to share that I'm moving to Boston to start my PhD at @Harvard! 🎓 Looking forward to new chapter and would love to connect with fellow researchers. Always happy to chat about AI, AI in healthcare, Robotics or grab coffee! #phd #AI #Boston #Harvard
8
492