Raschka's Top LLM Papers 2024 Sebastian Raschka's "best of 2024" on LLM research papers has some great resources: There's quite a broad selection of research themes, many quite interesting emerging areas such as parameter-efficient fine-tuning (e.g., LoRA, MoE), extending context windows, retrieval-augmented generation (RAG), multi-modal expansions that integrate visual or structured data, refined alignment strategies (like DPO and RLHF) to steer model outputs, knowledge editing for domain-specific modifications, and advanced compression (quantization, pruning). There's quite a lot of variety in this list, with also quite a lot of papers that delve into techniques for improved inference efficiency, dynamic architectures (state space models, mixture-of-experts), and novel prompting or training paradigms. If anyone is looking for a Christmas read, this is a great resource to catch up on 2024 LLM research over the winter holidays! |
|
|
|---|
|
DeepMind on Agent Cooperation DeepMind published an interesting paper exploring how LLM agents might learn to cooperate over multiple “generations” through an indirect reciprocity scenario: They tested three different models (Claude 3.5 Sonnet, Gemini 1.5 Flash, and GPT‑4o), some interesting insights such as showing that only Claude consistently evolved higher levels of cooperation across generations. Surprisingly, GPT‑4o tended to revert to "mutual defection", and Gemini 1.5 showed only "weak gains". It seems like there is a growing number of similar studies around LLMs for simulations of specific environments, which provide interesting sandboxes to test human-like interactions between actors at scale. |
|
|
|---|
|
Google's Globally Distributed DB An excellent paper for MLOps and Machine Learning Engineering practitioners is the classic Google’s Spanner paper: Google proposed an interesting globally distributed database architecture which offers strong external consistency at the cost of slight latency increase. The paper dives into how it integrates a two-phase commit with Paxos replication to automatically shard data and ensure high availability with lock-free snapshot reads. This commit wait is what introduces the design trade-off of slightly increased write latency (e.g., “commit wait”), but it is also what delivers synchronous replication and globally consistent reads, which is particularly useful for example in large-scale machine learning and MLOps systems requiring reproducible data snapshots, as well as the ability to evolve data schemas and indexes without downtime. |
|
|---|
|
Gentle Intro to Graph DL This is one of the best visual/intuitive introductions to Graph Neural Networks out there! As a reminder on GNNs, these graph neural networks are able to generalize deep learning by handling entities as nodes and their relationships as edges. This enables “passing messages” between connected nodes to learn context-aware representations that can power node-, edge-, or graph-level predictions. There are some nuances when working with GNNs vs traditional NNs, such as key design choices on how to represent nodes/edges, how to aggregate local or global information (e.g., via sum, mean, or max pooling), and how many layers to stack for broader context. GNNs are certainly worth learning, as a lot of real-world applications we see these relationships, and we have shown success in varied domains such as social networks, molecule property prediction, recommendation systems, etc. |
|
|
|---|
|
An Evolved Transformer Memory OpenAI O1's impressive performance at high compute ($3k est. per task) highlights that performance is the next tech race - Japanese startup Sakana AI brings a new approach to reduce memory by 75% with improved inference: Sakana AI releases this new “universal transformer memory” technique which uses small neural “attention memory” modules (NAMMs) to dynamically decide which tokens to keep or discard within an LLM’s context window. This sounds like quite a simple technique, but it promises to potentially cut memory usage by up to 75%. This is ceratinly going to be an interesting space to watch throughout 2025 and beyond! |
|
|
|---|
|
Upcoming MLOps Events The MLOps ecosystem continues to grow at break-neck speeds, making it ever harder for us as practitioners to stay up to date with relevant developments. A fantsatic way to keep on-top of relevant resources is through the great community and events that the MLOps and Production ML ecosystem offers. This is the reason why we have started curating a list of upcoming events in the space, which are outlined below. Upcoming conferences where we're speaking:
Other upcoming MLOps conferences in 2024:
In case you missed our talks:
|
|
|---|
| | |
Check out the fast-growing ecosystem of production ML tools & frameworks at the github repository which has reached over 10,000 ⭐ github stars. We are currently looking for more libraries to add - if you know of any that are not listed, please let us know or feel free to add a PR. Four featured libraries in the GPU acceleration space are outlined below. - Kompute - Blazing fast, lightweight and mobile phone-enabled GPU compute framework optimized for advanced data processing usecases.
- CuPy - An implementation of NumPy-compatible multi-dimensional array on CUDA. CuPy consists of the core multi-dimensional array class, cupy.ndarray, and many functions on it.
- Jax - Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
- CuDF - Built based on the Apache Arrow columnar memory format, cuDF is a GPU DataFrame library for loading, joining, aggregating, filtering, and otherwise manipulating data.
If you know of any open source and open community events that are not listed do give us a heads up so we can add them! |
|
|---|
| | |
As AI systems become more prevalent in society, we face bigger and tougher societal challenges. We have seen a large number of resources that aim to takle these challenges in the form of AI Guidelines, Principles, Ethics Frameworks, etc, however there are so many resources it is hard to navigate. Because of this we started an Open Source initiative that aims to map the ecosystem to make it simpler to navigate. You can find multiple principles in the repo - some examples include the following: - MLSecOps Top 10 Vulnerabilities - This is an initiative that aims to further the field of machine learning security by identifying the top 10 most common vulnerabiliites in the machine learning lifecycle as well as best practices.
- AI & Machine Learning 8 principles for Responsible ML - The Institute for Ethical AI & Machine Learning has put together 8 principles for responsible machine learning that are to be adopted by individuals and delivery teams designing, building and operating machine learning systems.
- An Evaluation of Guidelines - The Ethics of Ethics; A research paper that analyses multiple Ethics principles.
- ACM's Code of Ethics and Professional Conduct - This is the code of ethics that has been put together in 1992 by the Association for Computer Machinery and updated in 2018.
If you know of any guidelines that are not in the "Awesome AI Guidelines" list, please do give us a heads up or feel free to add a pull request!
|
|
|---|
| | |
| | | | The Institute for Ethical AI & Machine Learning is a European research centre that carries out world-class research into responsible machine learning. | | | | |
|
|
|---|
|
|
This email was sent to You received this email because you are registered with The Institute for Ethical AI & Machine Learning's newsletter "The Machine Learning Engineer"
|
| | | | |
|
|
|---|
|
© 2023 The Institute for Ethical AI & Machine Learning |
|
|---|
|
|
|
The Institute for Ethical AI & Machine Learning
