Gemma is expanding.... Google announced CodeGemma, a version of Gemma tuned for code generation. And bonus... Gemma is now bumped to v1.1, addressing lots of feedback we got.

Meta confirmed its GPT-4 competitor, Llama 3, is coming within the month.

At an event in London, Meta confirmed that it plans an initial release of Llama 3, its GPT-4 competitor, within the next month.

The company did not disclose the size of the parameters used in Llama 3, but it's expected to have about 140 billion parameters.

DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset

Distributed RObot Interaction Dataset: A diverse robot manipulation dataset with 76k demonstrations, collected across 564 scenes and 84 tasks over the course of a year.

Paper.

GE HealthCare’s Vscan Air SL with Caption AI software provides real-time guidance that shows healthcare professionals how to maneuver the probe to capture diagnostic-quality standard cardiac images.

With the help of on-device AI, there's now a way for handheld ultrasound users to confidently acquire cardiac views for rapid assessments at the point of care.

Meta announced 2nd-gen inference chip MTIAv2

- 708TF/s Int8 / 353TF/s BF16
- 256MB SRAM, 128GB memory
- 90W TDP. 24 chips per node, 3 nodes per rack.
- standard PyTorch stack (Dynamo, Inductor, Triton) for flexibility

Fabbed on TSMC's 5nm process, its fully programmable via the standard PyTorch stack, driven via Triton for software kernels.

This chip is an inference power-house and the software work is entirely driven by the PyTorch team, making usability a first; and its been great to see it in action on various Meta workloads.

New paper from Berkeley on Autonomous Evaluation and Refinement of Digital Agents

VLM/LLM-based evaluators can significantly improve the performance of agents for web browsing and device control, advancing sotas by 29% to 75%.

The music industry just had its 'ChatGPT for music' moment with Udio.

A new AI-powered music creation app called Udio from former Google DeepMind researchers just launched.

It allows users to generate full audio tracks in under 40 seconds with simple prompts and secured early funding from a16z, will i am, Common, and more.

A newly revealed patent from Microsoft Bing detailed ‘Visual Search’.

The patent describes a reverse image search with personal results tailored to user preferences and interests.

Integrated data visualization and analysis for workhorse biological assays

Make publication-quality figures in under a minute. No more fiddling with Excel, Prism, ggplot, and PowerPoint.

Google presents Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention

1B model that was fine-tuned on up to 5K sequence length passkey instances solves the 1M length problem.

The future of attention just dropped, and it looks a lot like a state space model (finite size, continual updates)

Little doubt now that a mixture of architectures will support the long-term, gradually conditioned memory needed for highly capable agents

Eye tracking as a window onto conscious and non-conscious processing in the brain.

The goals are two-fold and relatively simple:

1.      propose and test a method for familiarizing individuals with severe autism spectrum disorder (ASD) with the HoloLens 2 headset and the use of MR technology through a tutorial.

2.   obtain quantitative learning indicators in MR, such as execution speed and eye tracking, by comparing individuals with ASD to neurotypical individuals.

Over 80% of individuals with ASD successfully familiarized themselves with MR after several sessions.

In addition, the visual activity of individuals with ASD did not differ from that of neurotypical individuals when they successfully familiarized themselves.

This opens a lot of doors for potential learning opportunities in this population.

Apple nears production of first M4 chips with AI upgrades and plans to bring the new chips to all of its Macs, including new MacBook Pros and Airs, Mac Pro, Mac Studio, Mac mini and iMac across the end of this year and 2025.

Sanctuary AI announced a partnership with European automaker Magna.

The collab features Sanctuary AI’s development of general-purpose AI robots for deployment in Magna’s manufacturing operations.

Forbes released their AI 50 list.

Here’s a categorized list of the companies, including their current funding amounts.

Germany published the report "Generative AI Models - Opportunities and Risks for Industry and Authorities."

Quotes & comments:

1. “LLMs are trained based on huge text corpora. The origin of these texts and their quality are generally not fully verified due to the large amount of data. Therefore, personal or copyrighted data, as well as texts with questionable, false, or discriminatory content (e.g., disinformation, propaganda, or hate messages), may be included in the training set. When generating outputs, these contents may appear in these outputs either verbatim or slightly altered (Weidinger, et al., 2022). Imbalances in the training data can also lead to biases in the model" (page 9)

2. “If individual data points are disproportionately present in the training data, there is a risk that the model cannot adequately learn the desired data distribution and, depending on the extent, tends to produce repetitive, one-sided, or incoherent outputs (known as model collapse). It is expected that this problem will increasingly occur in the future, as LLM-generated data becomes more available on the internet and is used to train new LLMs (Shumailov, et al., 2023). This could lead to self-reinforcing effects, which is particularly critical in cases where texts with abuse potential have been generated, or when a bias in text data becomes entrenched. This happens, for example, as more and more relevant texts are produced and used again for training new models, which in turn generate a multitude of texts (Bender, et al., 2021)." (page 10)

3. “The high linguistic quality of the model outputs, combined with user-friendly access via APIs and the enormous flexibility of responses from currently popular LLMs, makes it easier for criminals to misuse the models for a targeted generation of misinformation (De Angelis, et al., 2023), propaganda texts, hate messages, product reviews, or posts for social media."

According to the report, special attention should be given to the following aspects:
➵ Raising awareness of users;
➵ Testing;
➵ Handling sensitive data;
➵ Establishing transparency;
➵ Auditing of inputs and outputs;
➵ Paying attention to (indirect) prompt injections;
➵ Selection and management of training data;
➵ Developing practical expertise.

Google DeepMind published new work on using reinforcement learning to train robots to be more agile (like soccer players).

They tested a new method of adding disruptive forces and randomness, which led to impressive results compared to the baseline.

Researchers at Tsinghua University in China have developed a revolutionary new AI chip that uses light instead of electricity to process data.

Dubbed “Taichi,” the chip is reportedly over 1,000 times more energy-efficient than Nvidia’s high performance H100 GPU chip.

The AI Index 2024 annual report by Stanford University is finally here.

This is the resource that will impact our conversations on the topic in the following months, highlighting the latest technical developments, its impact on global organizations, society and research, and the existential need for clear regulation and governance.

Executive exodus and a 10% cut to workforce at Tesla

Drew Baglino resigned from Tesla. Baglino had been at Tesla since 2006 and was SVP of Powertrain/Energy Engineering. He was one of just four named executive officers and integral to the work the company was doing on everything from EVs to energy storage and next gen 4680 cells. Drew posted on social media a few hours after our story to confirm his departure.

Tesla's Public Policy chief Rohan Patel also has left the company, we reported.

When you take in to account it follows key names in Tesla's semicondcutor team leaving earlier this year and Zach Kirkhorn leaving in August, its a lot of intellectual capital and experience out.

Actually things are going well - particularly in the energy division - and a reason for Baglino leaving is he felt the place was in good hands.

Last night Musk told staff around the world Tesla would cut 10% of more of its global workforce. The reasons are clear: pursuit of cost cuts and productivity in a tough environment. But Tesla's grown quickly. And Musk cited the number of duplicate roles in his reason for the RIF. 10% of Tesla's staff is around 14,000 people. So its sizable, the biggest RIF ever.

Paradigm announced an open source Reth Alphanet.

Reth AlphaNet is an OP Stack-compatible testnet rollup that maximizes Reth performance and enables experimentation of bleeding edge Ethereum Research.

Reth AlphaNet is a testnet rollup built on OP Stack & OP Reth.

Reth AlphaNet is aimed at experimentation of Ethereum research at the bleeding edge, and comes with 3 EIPs not available anywhere else yet. EIP-3074, EIP-71212, EIP-2537.

These EIPs are built with best-practices in mind, are optimized, and tested.

This is Microsoft Asia team overtaking original GPT-4 with their evol tuning

Today they are announcing WizardLM-2, next generation state-of-the-art LLM.

New family includes three cutting-edge models: WizardLM-2 8x22B, 70B, and 7B - demonstrates highly competitive performance compared to leading proprietary LLMs.

Model weights.