Decision-Making AI For The Enterprise

InstaDeep delivers AI-powered decision-making systems for the Enterprise. With expertise in both machine intelligence research and concrete business deployments, we provide a competitive advantage to our customers in an AI-first world.

Learn More

Building AI systems for the industry

Leveraging its expertise in GPU-accelerated computing, deep learning and reinforcement learning, InstaDeep has built AI systems to tackle the most complex challenges across a range of industries and sectors.

Biology Biology


Read More
Logistics Logistics


Read More
Electronic Design Electronic Design

Electronic Design

Read More
Energy Energy


Read More


Our latest updates from across our channels

InstaDeep at ICML 2024...

on Jul 18, 2024 | 02:04pm

The International Conference on Machine Learning (ICML) stands as one of the premier gatherings, bringing together the brightest minds from academia, industry, and research instit...

Syngenta and InstaDeep collaborate to accelerate crops seeds trai...

on Jun 18, 2024 | 01:30pm

This collaboration further strengthens the Syngenta Seeds R&D engine for speed, precision, and power, accelerating trait advancement.Large Language Models (LLMs) aim to reduce...

InstaDeep presents six papers at ICLR 2024...

on May 07, 2024 | 10:04am

InstaDeep maintains its strong commitment to open research with six papers accepted for presentation at the 2024 ICLR conference being held in Vienna this week. The accepted pa...

Building the next generation of AI models to decipher human biolo...

on Apr 30, 2024 | 03:41pm

The human genome, containing the entirety of our genetic blueprint, holds the keys to understand the intricate workings of our cells and bodies. How our cells respond to signals,...

From Finance to AI: Meet Divanisha Patel...

on Apr 04, 2024 | 01:16pm

Divanisha Patel was working for a bank when the possibilities AI holds for positive change caught her imagination. Divanisha decided to make the jump from finance to AI. Now a...

Diving into Deep Learning: Meet Narimane Hennouni...

on Mar 26, 2024 | 05:22pm

Can competitive swimming prepare you for the dive into AI? Once a competitive swimmer in national and international competitions for 18 years, Narimane Hennouni brings a c...

Mapping genomics with AI: Meet Maša Roller...

on Mar 22, 2024 | 05:25pm

Imagine waking to a world where every genetic sequence is a code waiting to be deciphered, pieces in a giant puzzle. That’s everyday life for Maša Roller, a senior computati...

TIME honours InstaDeep CEO for building a better world with AI...

on Feb 13, 2024 | 04:04pm

TIME Magazine awarded CEO and Co-Founder Karim Beguir a TIME100 Impact Award for leading InstaDeep's work to help communities and business tackle tough problems with AI. Ka...

InstaDeep CEO Karim Beguir Influential Africans

InstaDeep CEO on 100 Most Influential Africans list for second ye...

on Jan 15, 2024 | 02:53pm

InstaDeep CEO Karim Beguir has been named one of the 100 Most Influential Africans by a leading pan-African news magazine. The New African cited his leadership and innovation in A...

Tunis InstaDeep researchers at NeurIPS 2023’s NAML: North A...

on Dec 11, 2023 | 01:00pm

Two InstaDeep Tunis-based AI researchers will present at NeurIPS 2023 NAML: North African Machine Learning workshop on multi-script handwriting recognition and what happens Multi...


SMX: Sequential Monte Carlo Planning for Expert Iteration

Edan Toledo | Matthew Macfarlane | Donal John Byrne | Siddarth Singh | Paul Duckworth | Alexandre Laterre

ICML 2024 Jul 2024
Figure 1: Diagram depicting a representation of SMX search from left to right. N Rollouts are executed in parallel according to πθ (the sampling policy β). At each step in the environment the particle weights are adjusted, indicated by the particle sizes. We depict two resampling zones where particles are resampled (favouring higher weights) and weights are reset. Finally an improved policy π ′ = Iˆ βπ is constructed from the initial actions from the remaining particles, furthest to the right. This improved policy is then used to update πθ.

Multi-Objective Quality-Diversity for Crystal Structure Prediction

Hannah Janmohamed | Marta Wolinska | Shikha Surana | Aaron Walsh | Thomas Pierrot | Antoine Cully

Gecco 2024 Jul 2024

Overconfident Oracles: Limitations of In Silico Sequence Design Benchmarking

Shikha Surana | Nathan Grinsztajn | Timothy Atkinson | Paul Duckworth | Thomas D. Barrett

ICML 2024 workshop Jul 2024

Generative Model for Small Molecules with Latent Space RL Fine-Tuning to Protein Targets

Ulrich A. Mbou So | Qiulin Li | Dries Smit | Arnu Pretorius | Oliver Bent | Miguel Arbesú

ICML 2024 workshop Jul 2024
Figure 1. Schematic representation of our model’s architecture. A sequence of N tokens is passed as input to our encoder which is a transformer model. The output encoded embeddings of shape N × E are either passed directly to the mean and logvar layers (path 1) or they are first passed to the perceiver resampler layer which maps the encoded embeddings to a reduced dimension of shape LS ×LE (path 2). The mean and logvar layers are linear layers that are applied independently to each sequence dimension. The final reparametrised embeddings are then passed to the decoder transformer model to be used as encoder embeddings in the decoder’s cross-attention layers.

Should we be going MAD?
A Look at Multi-Agent Debate Strategies for LLMs

Andries Petrus Smit | Nathan Grinsztajn | Paul Duckworth | Thomas D Barrett | Arnu Pretorius

ICML 2024 Jul 2024
Recent advancements in large language models (LLMs) underscore their potential for responding to inquiries in various domains. However, ensuring that generative agents provide accurate and reliable answers remains an ongoing challenge. In this context, multi-agent debate (MAD) has emerged as a promising strategy for enhancing the truthfulness of LLMs. We benchmark a range of debating and prompting strategies to explore the trade-offs between cost, time, and accuracy. Importantly, we nd that multi-agent debating systems, in their current form, do not reliably outperform other proposed prompting strategies, such as self-consistency and ensembling using multiple reasoning paths. However, when performing hyperparameter tuning, several MAD systems, such as Multi-Persona, perform better. This suggests that MAD protocols might not be inherently worse than other approaches, but that they are more sensitive to different hyperparameter settings and difcult to optimize. We build on these results to offer insights into improving debating strategies, such as adjusting agent agreement levels, which can signicantly enhance performance and even surpass all other non-debate protocols we evaluated. We provide an open-source repository to the community with several state-of-theart protocols together with evaluation scripts to benchmark across popular research datasets.

Quality-Diversity for One-Shot Biological Sequence Design

Jérémie DONA | Arthur Flajolet | Andrei Marginean | Antoine Cully | Thomas PIERROT

ICML 20224 Jul 2024
Figure 1. Left. Schematic overview of our experimental protocol. An oracle, e.g. an expressive neural network is learned from real data. It enables us to relabel the dataset and emulates wet-lab results. An ensemble of scoring functions are learned from this relabelled dataset. Right. We optimize a MAP-ELITES grid with respect to this ensemble of scoring functions, following eq. (2), and the descriptors of eq. (4)

In the Press