Taiwan’s semiconductor industry has almost fully recovered from the 7.2-magntide earthquake that struck Wednesday, media report, citing the National Science and Technology Council (NSTC), which oversees Taiwan’s 3 big tech parks, Hsinchu Science Park (HSP), Central Taiwan Science Park (CTSP), and Southern Taiwan Science Park (STSP).

At HSP, the majority of semiconductor, display panel and other precision industries were back to normal on 4/3. Work continues at a small number of firms, but they are expected to be back to normal soon.
CTSP said 90% of affected semiconductor equipment is back online, and the rest will be back in operation today 4/4.
STSP said operations at all major factories are already back to normal.

AIDE has become the first human-level AI agent for data science

AIDE outperforms half of human data scientists on a wide range of Kaggle competitions, surpassing conventional AutoML, LangChain agents, and ChatGPT with human assistance.

On 4 April, Singapore effected changes to the Payment Services Act, expanding the scope of digital payment token (DPT) regulation in the country.

Here’s what those changes mean:

1. DPT service providers will now need to seek a licence in order to (i) provide custodial services for DPTs, or (ii) facilitate the transmission or exchange of DPTs, even where the service provider does not come into possession of client moneys or DPTs.

2. Businesses currently operating under the PSA’s expanded scope have 30 days to notify MAS of their activities, 6 months to submit a licence application, and 9 months to provide an attestation of their business activities and AML/CFT compliance by an external auditor.

3. Businesses that meet the above requirements can continue conducting business on a temporary basis while MAS reviews their licence applications.

4. In addition to the new licensing requirements, new consumer protection requirements that MAS finalised last year, such as on the safeguarding of customer assets, will come into force 6 months from 4 April.

Training LLMs can be much cheaper than previously thought.

While companies like OpenAI and Meta use billions of dollars to train theirs, CSAIL & Myshell research shows that just 0.1 million USD is sufficient for training LLaMA2-level LLMs.

JetMoE democratizes the training of high-performance LLMs, and makes it achievable by a wide range of research institutes and companies.

JetMoE is fully open-sourced & academia-friendly because:

1. It only uses public datasets for training. No proprietary resource is needed.
2. It can be finetuned with a very limited computing budget (e.g., consumer-grade GPU).

Photobucket is in talks with several AI companies to permit the use of its 13 billion photos and videos as training data.

Rates under discussion range from 5 cents to $1 per photo, and over $1 per video.

One prospective buyer told the CEO they want to buy over a billion videos.

OpenAI made a big upgrade to DALL-E 3, now allowing users to edit images directly in ChatGPT.

Users can edit images directly in the chat across the web, iOS, and Android apps by selecting an area of the image and prompting changes.

A super interesting talk on Ring Attention, probably the magic behind Gemini's 1 million context window

You organize your devices (GPU/TPU) in a ring, each computing a part of the final attention output

Each device needs to see all keys/values to produce its part. The idea is that the attention output can be computed blockwise (by splitting on the sequence dimension). Each device computes the updated queries of a chunk of the sequence by sending/receiving keys/values

This is a great repo to understand it in code.

Intelligent fabrics, which can sense and communicate information scalably and unobtrusively, can fundamentally change how people interact with the world.

Apple presents Ferret-UI

Grounded Mobile UI Understanding with Multimodal LLMs

Recent advancements in multimodal large language models (MLLMs) have been noteworthy, yet, these general-domain MLLMs often fall short in their ability to comprehend and interact effectively with user interface (UI) screens.

⚡️AutoCodeRover is autonomous software engineer from Singapore

Takes in a Github issue (bug fixing or feature addition), resolves in few minutes, with minimal LLM cost ~$0.5

Google released CodeGemma, a new version of the Gemma line of models fine-tuned on code generation and completion, that achieves state-of-the-art results. Available in sizes 2B and 7B.

HF is here.

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%.