AI Makerspace presents part 2 of the Alignment series, Aligning LLMs: RLAIF! They will tackle RLAIF, a technique that attempts to solve the fundamental bottleneck of RLHF, namely, that it’s kind of expensive to collect a ton of data directly from humans!
Accenture's latest Technology Vision report focuses on "human by design" technology. Contrasting the limits of non-human technology and the advent of generative AI as an inclusive, potentially revolutionary technology, the report identifies the trends shaping the future of technology.
South Korea-based startup, Rebellions AI closed its successful Series B with $124M; Figure AI is in talks with Microsoft and OpenAI; Eagle-7B is the strongest multi-language open-source LLM to date; The EU AI Act has practically become a law; and other relevant headlines for this week
AI Makerspace is hosting a deep dive into the OLMo series of models, including a discussion of the Dolma pre-training dataset as the model architectures leveraged for each of the three newly released versions: OLMo-1B, OLMo-7B, and OLMo-7B-Twin-2T!
AI2 is committed to releasing genuinely open-source models. This means that the model, the pre-training data, training code, model weights, training metrics, and all other data related to the models' development will be released under an Apache 2.0 license.
Semron's chips use capacitors with a memory component. The capacitor's heat-reducing capabilities enable the building of multiple layers of capacitors on a chip, yielding a reduced footprint. The memory component diminishes the noise in calculations, accounting for the chips' energetic efficiency.
Join AI Makerspace for a special event on RAG! Deep dive into the art of evaluating complex LLM applications that leverage Retrieval Augmented Generation (RAG), the technique that aims to ground the outputs of LLMs in fact-checkable information.
AlphaCodium is a code generation flow that improves the performance of LLMs without requiring fine-tuning. The flow can be used with closed and open-source models and was released with reproducible code and an evaluation script, enabling reproducible comparisons.
AI Makerspace is hosting a 4-part series on Aligning LLMs! In this first event on RLHF, they aim to give you some deep intuition into how models like InstructGPT and Llama 2 leveraged human feedback to align with us on what it means to be “helpful, honest, and harmless.”
Competera, a startup specializing in developing AI-powered price optimization services for retailers worldwide, recently closed a $3M seed funding round. Competera plans to use the funds to further its AI capabilities, expand its services, and keep up with demand, especially in the US market.
The Federal Trade Commission (FTC) issued 6(b) Orders to Alphabet, Amazon, Anthropic, Microsoft and OpenAI; Oracle finally revealed its Cloud Infrastructure Generative AI service; An MIT study concluded that AI is too expensive to replace human workers at the moment; MLflow 2.10.0, and more.
SKY ENGINE AI has just raised $7 million in a Series A funding round led by Cogito Capital Partners. The company is an industry leader in synthetic data generation for computer vision, serving customers such as Scania and Samsung.