MMAI Gym for Science: Training Liquid Foundation Models for Drug Discovery
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911483664793600 |
|---|---|
| author | Kuznetsov, Maksim Miftahutdinov, Zulfat Shayakhmetov, Rim Mizera, Mikolaj Schutski, Roman Zagribelnyy, Bogdan Ilin, Ivan Bondarev, Nikita MacDougall, Thomas Reymond, Mathieu Bafna, Mihir Kaymak-Loveless, Kaeli Babin, Eugene Malkov, Maxim Lechner, Mathias Hasani, Ramin Amini, Alexander Aladinskiy, Vladimir Aliper, Alex Zhavoronkov, Alex |
| author_facet | Kuznetsov, Maksim Miftahutdinov, Zulfat Shayakhmetov, Rim Mizera, Mikolaj Schutski, Roman Zagribelnyy, Bogdan Ilin, Ivan Bondarev, Nikita MacDougall, Thomas Reymond, Mathieu Bafna, Mihir Kaymak-Loveless, Kaeli Babin, Eugene Malkov, Maxim Lechner, Mathias Hasani, Ramin Amini, Alexander Aladinskiy, Vladimir Aliper, Alex Zhavoronkov, Alex |
| contents | General-purpose large language models (LLMs) that rely on in-context learning do not reliably deliver the scientific understanding and performance required for drug discovery tasks. Simply increasing model size or introducing reasoning tokens does not yield significant performance gains. To address this gap, we introduce the MMAI Gym for Science, a one-stop shop molecular data formats and modalities as well as task-specific reasoning, training, and benchmarking recipes designed to teach foundation models the 'language of molecules' in order to solve practical drug discovery problems. We use MMAI Gym to train an efficient Liquid Foundation Model (LFM) for these applications, demonstrating that smaller, purpose-trained foundation models can outperform substantially larger general-purpose or specialist models on molecular benchmarks. Across essential drug discovery tasks - including molecular optimization, ADMET property prediction, retrosynthesis, drug-target activity prediction, and functional group reasoning - the resulting model achieves near specialist-level performance and, in the majority of settings, surpasses larger models, while remaining more efficient and broadly applicable in the domain. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_03517 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MMAI Gym for Science: Training Liquid Foundation Models for Drug Discovery Kuznetsov, Maksim Miftahutdinov, Zulfat Shayakhmetov, Rim Mizera, Mikolaj Schutski, Roman Zagribelnyy, Bogdan Ilin, Ivan Bondarev, Nikita MacDougall, Thomas Reymond, Mathieu Bafna, Mihir Kaymak-Loveless, Kaeli Babin, Eugene Malkov, Maxim Lechner, Mathias Hasani, Ramin Amini, Alexander Aladinskiy, Vladimir Aliper, Alex Zhavoronkov, Alex Machine Learning Artificial Intelligence Computation and Language General-purpose large language models (LLMs) that rely on in-context learning do not reliably deliver the scientific understanding and performance required for drug discovery tasks. Simply increasing model size or introducing reasoning tokens does not yield significant performance gains. To address this gap, we introduce the MMAI Gym for Science, a one-stop shop molecular data formats and modalities as well as task-specific reasoning, training, and benchmarking recipes designed to teach foundation models the 'language of molecules' in order to solve practical drug discovery problems. We use MMAI Gym to train an efficient Liquid Foundation Model (LFM) for these applications, demonstrating that smaller, purpose-trained foundation models can outperform substantially larger general-purpose or specialist models on molecular benchmarks. Across essential drug discovery tasks - including molecular optimization, ADMET property prediction, retrosynthesis, drug-target activity prediction, and functional group reasoning - the resulting model achieves near specialist-level performance and, in the majority of settings, surpasses larger models, while remaining more efficient and broadly applicable in the domain. |
| title | MMAI Gym for Science: Training Liquid Foundation Models for Drug Discovery |
| topic | Machine Learning Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2603.03517 |