Christopher Abelt

Digital Marketer. Growth Strategist. Big on Sustainable Businesses.

[Christopher Abelt @cabelt]

https://t.co/VucC5qS17M

[Christopher Abelt @cabelt]

Check out my latest article: In a world of infinite content, only simple stories win. https://t.co/Nsosn96P6v via @LinkedIn

[Christopher Abelt @cabelt]

Check out my latest article: Barriers to Growth for RIAs – And How to Overcome Them https://t.co/T8FKdn3SLu via @LinkedIn

[Christopher Abelt @cabelt]

Attending @economistimpact Sustainability Week event in NYC featuring @ChelseaClinton, @VijayVaitheesw3 and Ann Lee from CORE on strengthening #climate resilience

[Christopher Abelt @cabelt]

Check out my latest article: Why Should Earth Day Matter Beyond April? https://t.co/5BHaOaOKib via @LinkedIn

[Christopher Abelt @cabelt]

https://t.co/ESebwWBqAn

[New York Rangers @NYRangers]

This team.

[Jim Fan @DrJimFan]

I also recommend reading @natolambert's excellent blog post that summarizes the key points, if you don't have time for 36 pages of whitepaper: https://t.co/zGcesFlzha

[Jim Fan @DrJimFan]

You'll soon see lots of "Llama just dethroned ChatGPT" or "OpenAI is so done" posts on Twitter. Before your timeline gets flooded, I'll share my notes:

▸ Llama-2 likely costs $20M+ to train. Meta has done an incredible service to the community by releasing the model with a commercially-friendly license. AI researchers from big companies were wary of Llama-1 due to licensing issues, but now I think many of them will jump on the ship and contribute their firepower.

▸ Meta's team did a human study on 4K prompts to evaluate Llama-2's helpfulness. They use "win rate" as a metric to compare models, in similar spirit as the Vicuna benchmark. 70B model roughly ties with GPT-3.5-0301, and performs noticeably stronger than Falcon, MPT, and Vicuna.
I trust these real human ratings more than academic benchmarks, because they typically capture the "in-the-wild vibe" better.

▸ Llama-2 is NOT yet at GPT-3.5 level, mainly because of its weak coding abilities. On "HumanEval" (standard coding benchmark), it isn't nearly as good as StarCoder or many other models specifically designed for coding. That being said, I have little doubt that Llama-2 will improve significantly thanks to its open weights.

▸ Meta's team goes above and beyond on AI safety issues. In fact, almost half of the paper is talking about safety guardrails, red-teaming, and evaluations. A round of applause for such responsible efforts!

In prior works, there's a thorny tradeoff between helpfulness and safety. Meta mitigates this by training 2 separate reward models. They aren't open-source yet, but would be extremely valuable to the community.

▸ I think Llama-2 will dramatically boost multimodal AI and robotics research. These fields need more than just blackbox access to an API.
So far, we have to convert the complex sensory signals (video, audio, 3D perception) to text description and then feed to an LLM, which is awkward and leads to huge information loss. It'd be much more effective to graft sensory modules directly on a strong LLM backbone.

▸ The whitepaper itself is a masterpiece. Unlike GPT-4's paper that shared very little info, Llama-2 spelled out the entire recipe, including model details, training stages, hardware, data pipeline, and annotation process. For example, there's a systematic analysis on the effect of RLHF with nice visualizations.

Quote sec 5.1: "We posit that the superior writing abilities of LLMs, as manifested in surpassing human annotators in certain tasks, are fundamentally driven by RLHF."

Congrats to the team again 🥂! Today is another delightful day in OSS AI.

[Physis Investment @physisinvest]

Two recent reports on climate change and related financial, economic, and social risks should be required reading for investors and other stakeholders as it will take all hands on deck to offset record climate change numbers. Learn more in our latest blog: https://t.co/YHMq539I4Y

[Christopher Abelt @cabelt]

https://t.co/wfC8tcxUYa

[Christopher Abelt @cabelt]

What is US #energysecurity ? Why does it matter? https://t.co/NrfCFAUYvF via @LinkedIn

[Christopher Abelt @cabelt]

Check out my latest article: What is Permitting Reform? Can it Help the US Energy Sector Reduce its Carbon Footprint? https://t.co/Z9oVESErdo via @LinkedIn

[Christopher Abelt @cabelt]

Check out my latest article: How Can Artificial Intelligence Help the Transition to Global Green Energy? https://t.co/CarXB5GRy9 via @LinkedIn

[Christopher Abelt @cabelt]

Ad industry meme accounts have evolved from sharing crude jokes to breaking news and helping marketers find jobs https://t.co/iJu0pVYWa9 via @WSJ

[BBC News (World) @BBCWorld]

It's 50 years since Silent Spring by Rachel Carson was published - inspiration for the Green movement. @BBCWitness: http://t.co/i4RcPLRA

[Christopher Abelt @cabelt]

Check out my latest article: To B or Not to B. A Question Relevant to ESG for 2023? https://t.co/5yYStRkUAK via @LinkedIn

[Christopher Abelt @cabelt]

https://t.co/UVdbDe5xAa

[Christopher Abelt @cabelt]

As Climate Protests Get Bolder, British Police Strike Back With New Powers https://t.co/9EoIm89sTM

[Christopher Abelt @cabelt]

War in Ukraine Likely to Speed, Not Slow, Shift to Clean Energy, @IEA says https://t.co/hiCv9kHg1k