Jsevillamol

LESSWRONG
LW

Jsevillamol — LessWrong

Introducing the Epoch Capabilities Index (ECI)

luke_emberson, YafahEdelman, Jsevillamol

4mo

We at Epoch AI have recently released a new composite AI capability index called the Epoch Capabilities Index (ECI), based on nearly 40 underlying benchmarks.

Some key features...

Saturation-proof: ECI "stitches" benchmarks together, to enable comparisons even as individual benchmarks become saturated.
Global comparisons: Models can be compared, even if they were never evaluated on the same benchmarks.
Difficulty-based task weighting: ECI uses a simple statistical model (similar to those used in Item Response Theory) under which models deemed more capable if they score well on difficult benchmarks, and benchmarks are deemed more difficult if capable models are unable to score highly on them.

ECI will allow us to track trends in capabilities over longer spans of... (read more)

Power laws in Speedrunning and Machine Learning

Jsevillamol

Jsevillamol, Ege Erdil

Anticipating the future of AI necessarily requires anticipating results in performance. Ideally, we would like to understand the dynamics of improvements. This would help us preempt capabilities and understand how plausible sudden improvements are.

However, this problem is notoriously difficult. Among other reasons, Machine Learning benchmarks use many different metrics for measuring performance. And the history of improvements in all domains is limited, spanning around a dozen improvements in the longest-running benchmarks.

To circumvent these problems, we follow Sevilla (2021) and study video game speedrunning. Using data from speedrun.com, we investigate a previously noted regularity in world record progressions - an astounding fit to a power law pattern.

... (read 284 more words →)

Announcing Epoch’s dashboard of key trends and figures in Machine Learning

Jsevillamol

Developments in Machine Learning have been happening extraordinarily fast, and as their impacts become increasingly visible, it becomes ever more important to develop a quantitative understanding of these changes. However, relevant data has thus far been scattered across multiple papers, has required expertise to gather accurately, or has been otherwise hard to obtain.

Given this, Epoch is thrilled to announce the launch of our new dashboard, which covers key numbers and figures from our research to help understand the present and future of Machine Learning. This includes:

Training compute requirements
Model size, measured by the number of trainable parameters
The availability and use of data for training
Trends in hardware efficiency
Algorithmic improvements for achieving better performance with

... (read 168 more words →)

Epoch Impact Report 2022

Jsevillamol

Epoch is a research group forecasting the development of transformative Artificial Intelligence. We try to understand how progress in AI happens and what economic impacts we might see from advanced AI.

We want to enable better governance during this economic transition by gathering information about the timing of new developments, studying which levers can be used to influence AI progress and making current and past trends in ML more understandable.

Founded in April of 2022, Epoch currently has a staff of 13 people, corresponding to 9 FTEs. We have received $1.96M in funding through a grant from Open Philanthropy. We are fiscally sponsored and operationally supported by Rethink Priorities, whose Special Projects team has... (read more)

Literature review of TAI timelines

Jsevillamol

Jsevillamol, keith_wynroe, David Atkinson

We summarize and compare several models and forecasts predicting when transformative AI will be developed.

Highlights

The review includes quantitative models, including both outside and inside view, and judgment-based forecasts by (teams of) experts.
While we do not necessarily endorse their conclusions, the inside-view model the Epoch team found most compelling is Ajeya Cotra’s “Forecasting TAI with biological anchors”, the best-rated outside-view model was Tom Davidson’s “Semi-informative priors over AI timelines”, and the best-rated judgment-based forecast was Samotsvety’s AGI Timelines Forecast.
The inside-view models we reviewed predicted shorter timelines (e.g. bioanchors has a median of 2052) while the outside-view models predicted longer timelines (e.g. semi-informative priors has a median over 2100). The judgment-based forecasts are skewed

... (read 452 more words →)

Injecting some numbers into the AGI debate - by Boaz Barak

Jsevillamol

I found this article useful context for Boaz Barak's more recent article about the probability of AI Doom.

In this article, the author (through very rough BOTECs):

Estimates the number of parameters needed for AGI (tentative conclusion: x10-100 more than GPT-3)
Estimates the cost of training AGI of that scale (tentative conclusion: $50B - $5T)
Estimates the cost of inference for AGI of that scale (tentative conclusion: $10/hour - $100/hour)
Notes some challenges that will need to be addressed to reach AGI
In the end there is a reflection on consciousness and AI risk

Here are some excerpts:

Bottom line. Overall, it seems that [we would need to scale model size by] at least 10-100 [to reach human-level AGI], though

... (read 713 more words →)

AI Forecasting Research Ideas

Jsevillamol

Overview

The linked document contains a collection of AI Forecasting research ideas, prepared by some Epoch employees in a personal capacity
We think that these are interesting and valuable projects that research interns or students could look into, though they may vary in difficulty (depending on your background/experience)
This is the result of a quick brainstorming and curation, rather than a thorough deliberative process. We encourage a critical outlook when reading them
You may also be interested in these other forecasting research ideas suggested by Jaime Sevilla
You can use Epoch’s database as a resource for finding notable machine learning papers with parameter, compute, and dataset sizes. On Epoch website you can also find our past research, a tool for visualising the dataset, and some other tools like this compute calculator.
Please feel free to contact us for clarification about these questions!

Some research ideas in forecasting

Jsevillamol

In the past, I have researched how we can effectively pool the predictions of many experts. For the most part, I am now focusing on directing Epoch and AI forecasting.

However, I have accumulated a log of research projects related to forecasting. I have the vague intention of working on them at some point, but this will likely be months or years. Meanwhile, I would be elated if someone else took my ideas and developed them.

And with the Million Predictions Hackathon by Metaculus looming, now seems a particularly good moment to write down some of these project ideas.

Compare different aggregation methods

Difficulty: easy

The ultimate arbiter of what aggregation works is what performs best in... (read 1748 more words →)

ACX Meetup Madrid

Pablo Villalobos

Pablo Villalobos, Jsevillamol

As part of the ACX Meetups Everywhere 2022, we'll be hosting an ACX meetup!

This is the exact location. There will be an ACX sign to make us easy to find. Feel free to come or leave at any point during the meetup.

The longest training run

Jsevillamol

Jsevillamol, Tamay, Owen D, anson.ho

In short: Training runs of large Machine Learning systems are likely to last less than 14-15 months. This is because longer runs will be outcompeted by runs that start later and therefore use better hardware and better algorithms. [Edited 2022/09/22 to fix an error in the hardware improvements + rising investments calculation]

Scenario	Longest training run
Hardware improvements	3.55 years
Hardware improvements + Software improvements	1.22 years
Hardware improvements + Rising investments	9.12 months
Hardware improvements + Rising investments + Software improvements	2.52 months

Larger compute budgets and a better understanding of how to effectively use compute (through, for example, using scaling laws) are two major driving forces of progress in recent Machine Learning.

There are many ways to increase your effective compute budget:... (read 2657 more words →)

LESSWRONG
LW

LESSWRONG
LW

Announcing Epoch: A research organization investigating the road to Transformative AI

Compute Trends Across Three eras of Machine Learning

A Bayesian Aggregation Paradox

A time-invariant version of Laplace's rule

Jsevillamol

Jsevillamol

Introducing the Epoch Capabilities Index (ECI)

Power laws in Speedrunning and Machine Learning

Announcing Epoch’s dashboard of key trends and figures in Machine Learning

Epoch Impact Report 2022

Literature review of TAI timelines

Injecting some numbers into the AGI debate - by Boaz Barak

AI Forecasting Research Ideas

Trends in Machine Learning

Jsevillamol

Announcing Epoch: A research organization investigating the road to Transformative AI

Compute Trends Across Three eras of Machine Learning

A Bayesian Aggregation Paradox

A time-invariant version of Laplace's rule

Jsevillamol

Jsevillamol

Introducing the Epoch Capabilities Index (ECI)

Power laws in Speedrunning and Machine Learning

Announcing Epoch’s dashboard of key trends and figures in Machine Learning

Epoch Impact Report 2022

Literature review of TAI timelines

Injecting some numbers into the AGI debate - by Boaz Barak

AI Forecasting Research Ideas

Trends in Machine Learning

Highlights