Deep double descent

OpenAI Blog2019-12-05 16:000 views

We show that the double descent phenomenon occurs in CNNs, ResNets, and transformers: performance first improves, then gets worse, and then improves again with increasing model size, data size, or training time. This effect is often avoided through careful regularization. While this behavior appears to be fairly universal, we don’t yet fully understand why it happens, and view further study of this phenomenon as an important research direction.

Source：https://openai.com/index/deep-double-descent

My favorites

PreviousDota 2 with large scale deep reinforcement learning NextProcgen Benchmark

An AI agent with no step limit stops when its context window is full. On GPT-5.6 that's $160 for one run.

The first agent most of us write is a while True loop. Ask the model what to do, run the tool it picks, append the resul

0dev.to

Phân tích biến động nhân sự Liverpool: Phương án Pierre Ismaelo Ganiou và bài toán tái thiết hàng thủ

Trước bối cảnh Virgil van Dijk bước vào năm cuối hợp đồng và giải pháp Ronald Araujo chỉ mang tính ngắn hạn, ban huấn lu

0dev.to Data

SpaceX Starship Flight 15: the ship was caught, the booster was not

Originally published on ColdCaseTech . Disclosure: this post contains affiliate links. We may earn a commission at no ex

0dev.to News

Two challenges: build the Notion or Trello replacement with your AI assistant ($500 each)

AI agents can write software now. Until now, nothing was built for them to run it. An agent can write a wiki or a task b

0dev.to Challenge

An overdue task isn't evidence that it happened

My reminder cleanup surfaced 30 open rows with a past due date after the first pass. That told me the list needed attent

0dev.to Discuss

A Superwhisper-style alternative on Windows — if you need Chinese and English, and no monthly subscription

If you like Superwhisper’s “press a key, speak, text lands at the cursor” idea, the usual follow-up is: what do you use

0dev.to Productivity