CARVIEW

MOTORHOMES

Select Language

HTTP/2 200 server: GitHub.com content-type: text/html; charset=utf-8 last-modified: Wed, 03 Dec 2025 13:58:42 GMT access-control-allow-origin: * etag: W/"69304212-36ca" expires: Mon, 29 Dec 2025 08:09:24 GMT cache-control: max-age=600 content-encoding: gzip x-proxy-cache: MISS x-github-request-id: 3C49:3ABDEF:86CDD2:9796A7:695234DA accept-ranges: bytes age: 0 date: Mon, 29 Dec 2025 07:59:24 GMT via: 1.1 varnish x-served-by: cache-bom-vanm7210033-BOM x-cache: MISS x-cache-hits: 0 x-timer: S1766995164.978024,VS0,VE202 vary: Accept-Encoding x-fastly-request-id: fccc23031ccbaea90f592f34c9fd18b4be40bf8f content-length: 3575 ICBINB - Crack open the research process

ICBINB: Crack open the research process

News

[Dec/2025] The next ICBINB workshop takes place at ICLR 2026! This year's theme: Where large language models need to improve.
[Dec/2024] We are organizing the next ICBINB workshop at ICLR 2025! We will take a deep into the pitfalls and challenges of applied deep learning.
[Jun/2023] The ICBINB workshop will be back at NeurIPS 2023! This time with a focus on failure modes of foundation models.
[Jan/2023] The talks of our NeurIPS 2022 workshop are now online. Watch them here!

What is ICBINB?

The ICBINB initiative is a movement within the ML community for well-executed meaningful research beyond bold numbers. The goals of the initiative are to crack open the research process, to re-value unexpected negative results, question well-established default practices, and advance the understanding, elegance, and diversity of the field as opposed to focusing solely on the outcome and just rewarding approaches that beat previous works on a given benchmark.

Objectives

The three pillars of such an initiative include:

Shed light on the research process: sharing stories about how research is really done behind the curtains, encouraging transparency and reproducibility.
Provide a platform for high-quality but under-valued ML research: showcase, disseminate, and support valuable work that is currently under-represented given publication incentives. This includes negative results/failed attempts, simple approaches that work well in practice, and applied work.
Develop an inclusive and welcoming community of researchers and practitioners that share the same values to support and help each other to conduct deep high-impact research. Encourage meta-dialog on how we should be conducting top-tier ML research.