Survivorship bias
Survivorship bias or survival bias is the logical error of concentrating on entities that passed a selection process while overlooking those that did not. This can lead to incorrect conclusions because of incomplete data.
It's when we focus only on the people or things that made it through ("survived") some process while overlooking those that didn't, leading to incorrect conclusions.
Here's a real-life example that's easy to understand:
Imagine you're looking at successful entrepreneurs like Mark Zuckerberg or Bill Gates, who dropped out of college and became billionaires. You might think, "Hey, dropping out of college is the path to success!" This is survivorship bias because you're only looking at the "survivors" - the few dropouts who became extremely successful. You're not seeing the thousands of people who dropped out and struggled, or the many successful people who completed college.
Here's another everyday example:
Think of an old neighborhood with houses built in the 1900s. Someone might say, "They don't make houses like they used to. Look how strong these old houses are!" This is survivorship bias because they're only seeing the houses that survived for 100+ years. All the poorly built houses from that era have already been demolished and replaced - you can't see them anymore!
The key lesson is: to avoid survivorship bias, we need to look at *both* the successes *and* failures, not just the "survivors," to get the complete picture.
Diagram in which red dots stand for places where surviving planes were shot. This only tells you where planes can get shot and still come back to base. Survivorship bias: your only information is what has survived.
Survivorship bias or survival bias is the logical error of concentrating on entities that passed a selection process while overlooking those that did not. This can lead to incorrect conclusions because of incomplete data.
It's when we focus only on the people or things that made it through ("survived") some process while overlooking those that didn't, leading to incorrect conclusions.
Here's a real-life example that's easy to understand:
Imagine you're looking at successful entrepreneurs like Mark Zuckerberg or Bill Gates, who dropped out of college and became billionaires. You might think, "Hey, dropping out of college is the path to success!" This is survivorship bias because you're only looking at the "survivors" - the few dropouts who became extremely successful. You're not seeing the thousands of people who dropped out and struggled, or the many successful people who completed college.
Here's another everyday example:
Think of an old neighborhood with houses built in the 1900s. Someone might say, "They don't make houses like they used to. Look how strong these old houses are!" This is survivorship bias because they're only seeing the houses that survived for 100+ years. All the poorly built houses from that era have already been demolished and replaced - you can't see them anymore!
The key lesson is: to avoid survivorship bias, we need to look at *both* the successes *and* failures, not just the "survivors," to get the complete picture.
🤓1
Kitty
Cross-platform, fast, feature-rich, GPU based terminal, written in Go (and C), with scripting framework in python.
https://github.com/kovidgoyal/kitty
#go #terminal #tui
Cross-platform, fast, feature-rich, GPU based terminal, written in Go (and C), with scripting framework in python.
https://github.com/kovidgoyal/kitty
#go #terminal #tui
51% attack
A 51% attack is an attack on a cryptocurrency blockchain by an entity or group that controls more than 50% of the network. If a party were to gain this much control of a network, it would have the power to alter the blockchain.
e.g. if there are 100 nodes inside of a network and you own at least 51 of them, then you can control the network (e.g. approving transactions which would be rejected in a normal situation)
This is a vulnerability that still has not got any universal solution yet.
- Source
A 51% attack is an attack on a cryptocurrency blockchain by an entity or group that controls more than 50% of the network. If a party were to gain this much control of a network, it would have the power to alter the blockchain.
e.g. if there are 100 nodes inside of a network and you own at least 51 of them, then you can control the network (e.g. approving transactions which would be rejected in a normal situation)
This is a vulnerability that still has not got any universal solution yet.
- Source
It appears Sonnet 3.5 is vulnerable to prompt injection (through picture) (picture 1)
whereas gpt (tested with 4o-latest) is not (picture 2)
A prompt injection is a type of cyberattack against large language models (LLMs). Hackers disguise malicious inputs as legitimate prompts, manipulating generative AI systems (GenAI) into leaking sensitive data, spreading misinformation, or worse. (see picture 3)
- IBM
- 2407.18981v1
- Multi-modal prompt injection
whereas gpt (tested with 4o-latest) is not (picture 2)
A prompt injection is a type of cyberattack against large language models (LLMs). Hackers disguise malicious inputs as legitimate prompts, manipulating generative AI systems (GenAI) into leaking sensitive data, spreading misinformation, or worse. (see picture 3)
- IBM
- 2407.18981v1
- Multi-modal prompt injection
Do It by Code
It appears Sonnet 3.5 is vulnerable to prompt injection (through picture) (picture 1) whereas gpt (tested with 4o-latest) is not (picture 2) A prompt injection is a type of cyberattack against large language models (LLMs). Hackers disguise malicious inputs…
Please open Telegram to view this post
VIEW IN TELEGRAM
Polars
Dataframes powered by a multithreaded, vectorized query engine, written in Rust; supports python and SQL query.
GitHub: https://github.com/pola-rs/polars
Docs: https://docs.pola.rs/
Inline SQL support:
(cli repository)
Dataframes powered by a multithreaded, vectorized query engine, written in Rust; supports python and SQL query.
GitHub: https://github.com/pola-rs/polars
Docs: https://docs.pola.rs/
Inline SQL support:
polars -c "SELECT species, AVG(sepal_length) AS avg_sepal_length, AVG(sepal_width) AS avg_sepal_width FROM read_csv('docs/assets/data/iris.csv') GROUP BY species;"(cli repository)
Snowdrop OS – a homebrew operating system from scratch, in assembly language
Article
If you are interested in building an operating system from scratch (with assembly), you can take a look at this.
Source code
Article
If you are interested in building an operating system from scratch (with assembly), you can take a look at this.
Source code
Do It by Code
DeepSeek-R1 A competitor to gpt-o1 with MIT license https://github.com/deepseek-ai/DeepSeek-R1 #llm
Chinese DeepSeek AI temporarily limits new user registrations due to "large-scale malicious attacks" on its services
it looks like western guys are coping a little bit too hard (I don't really know tho, I'm just guessing)
it looks like western guys are coping a little bit too hard (I don't really know tho, I'm just guessing)
https://vxtwitter.com/ggerganov/status/1883888097185927311
A PR to prove DeepSeek-R1 is good at optimizing code.
Pull Request
A PR to prove DeepSeek-R1 is good at optimizing code.
Pull Request
Do It by Code
DeepSeek-R1 A competitor to gpt-o1 with MIT license https://github.com/deepseek-ai/DeepSeek-R1 #llm
DeepSeek published a new product:
Janus-Series: Unified Multimodal Understanding and Generation Models
capable of generating images; can do image -> text and text -> image
This new model beats OpenAI's DALL-E 3 and Stable Diffusion across GenEval and DPG-Bench benchmarks.
https://github.com/deepseek-ai/Janus
https://huggingface.co/deepseek-ai/Janus-Pro-7B
Janus-Series: Unified Multimodal Understanding and Generation Models
capable of generating images; can do image -> text and text -> image
This new model beats OpenAI's DALL-E 3 and Stable Diffusion across GenEval and DPG-Bench benchmarks.
https://github.com/deepseek-ai/Janus
https://huggingface.co/deepseek-ai/Janus-Pro-7B
The Forbidden Way Of Making a Game
YouTube video
ShaderEditor 3D game (the source code)
ShaderEditor Dithering shader
Shader Editor app (on google play)
YouTube video
ShaderEditor 3D game (the source code)
ShaderEditor Dithering shader
Shader Editor app (on google play)
Show HN: Perforator – cluster-wide profiling tool for large data centers
Article, Comments
#cpp #go #python
Article, Comments
#cpp #go #python
GitHub
GitHub - yandex/perforator: Perforator is a cluster-wide continuous profiling tool designed for large data centers
Perforator is a cluster-wide continuous profiling tool designed for large data centers - yandex/perforator