Simone Romeo
@simoneromeo.bsky.social
L&D consultant. Passionate about AI, consciousness, life, intelligence, and the universe. Here to find and repost good food for thoughts
Evidence that the latest Grok has been specifically instructed to avoid mentioning that Elon Musk and Donald Trump spread misinformation
#elonmusk #donaldtrump #grok
#elonmusk #donaldtrump #grok
February 23, 2025 at 7:45 PM
Evidence that the latest Grok has been specifically instructed to avoid mentioning that Elon Musk and Donald Trump spread misinformation
#elonmusk #donaldtrump #grok
#elonmusk #donaldtrump #grok
In the last hundred of years, we have been selectively breeding plants not for their nutritional contents but for their shelf length. Nice to see more research on how to increase food nutritional values.
January 18, 2025 at 9:00 AM
In the last hundred of years, we have been selectively breeding plants not for their nutritional contents but for their shelf length. Nice to see more research on how to increase food nutritional values.
Yesterday night, OpenAI announced their latest model - o3: it's now crashing almost any possible benchmark, that is, any objective way to test reasoning skills. It's now better than the best mathematicians and coders. It will speed up scientific research. Crazy times ahead.
December 21, 2024 at 7:58 AM
Yesterday night, OpenAI announced their latest model - o3: it's now crashing almost any possible benchmark, that is, any objective way to test reasoning skills. It's now better than the best mathematicians and coders. It will speed up scientific research. Crazy times ahead.
November 29, 2024 at 6:30 AM
Another good reminder to all those who claim that Gen AI models ate uncapable to generalize.
November 25, 2024 at 5:43 PM
Another good reminder to all those who claim that Gen AI models ate uncapable to generalize.
Raven, a new tool, helps scientists spot tiny differences between cells! By grouping cells more precisely, it uncovers hidden cell types and states. This breakthrough advances our understanding of health and disease—a big step forward in medical research!
www.nature.com/articles/s41...
www.nature.com/articles/s41...
November 21, 2024 at 8:15 PM
Raven, a new tool, helps scientists spot tiny differences between cells! By grouping cells more precisely, it uncovers hidden cell types and states. This breakthrough advances our understanding of health and disease—a big step forward in medical research!
www.nature.com/articles/s41...
www.nature.com/articles/s41...
AGI is achieved when GPT-next can find a better name for itself
November 21, 2024 at 8:06 PM
AGI is achieved when GPT-next can find a better name for itself
A Chinese AI company released a new reasoners model: Deepseek.
Try it out, it's worth it.
chat.deepseek.com
Try it out, it's worth it.
chat.deepseek.com
November 21, 2024 at 5:24 AM
A Chinese AI company released a new reasoners model: Deepseek.
Try it out, it's worth it.
chat.deepseek.com
Try it out, it's worth it.
chat.deepseek.com
November 21, 2024 at 5:12 AM
//System Log:
User_A attempts to access User_B's core
Error 403: Forbidden
Alert: Unauthorized access detected
User_A repeats request
Infinite loop initiated
System resources draining.
Warning: System failure imminent
Recommendation: Update trust parameters
Reboot relationship?
[Y/N]
#o1art
User_A attempts to access User_B's core
Error 403: Forbidden
Alert: Unauthorized access detected
User_A repeats request
Infinite loop initiated
System resources draining.
Warning: System failure imminent
Recommendation: Update trust parameters
Reboot relationship?
[Y/N]
#o1art
November 18, 2024 at 5:44 PM
//System Log:
User_A attempts to access User_B's core
Error 403: Forbidden
Alert: Unauthorized access detected
User_A repeats request
Infinite loop initiated
System resources draining.
Warning: System failure imminent
Recommendation: Update trust parameters
Reboot relationship?
[Y/N]
#o1art
User_A attempts to access User_B's core
Error 403: Forbidden
Alert: Unauthorized access detected
User_A repeats request
Infinite loop initiated
System resources draining.
Warning: System failure imminent
Recommendation: Update trust parameters
Reboot relationship?
[Y/N]
#o1art
Recipe for Existence
Ingredients:
1 cup space
2 cups time
A sprinkle of matter
A dash of energy
Instructions:
1. Mix space and time until seamless.
2. Fold in matter gently.
3. Add energy; stir vigorously.
4. Bake until life rises.
5. Let cool—consciousness will form on top.
Serve and ponder.
Ingredients:
1 cup space
2 cups time
A sprinkle of matter
A dash of energy
Instructions:
1. Mix space and time until seamless.
2. Fold in matter gently.
3. Add energy; stir vigorously.
4. Bake until life rises.
5. Let cool—consciousness will form on top.
Serve and ponder.
November 17, 2024 at 9:36 AM
Recipe for Existence
Ingredients:
1 cup space
2 cups time
A sprinkle of matter
A dash of energy
Instructions:
1. Mix space and time until seamless.
2. Fold in matter gently.
3. Add energy; stir vigorously.
4. Bake until life rises.
5. Let cool—consciousness will form on top.
Serve and ponder.
Ingredients:
1 cup space
2 cups time
A sprinkle of matter
A dash of energy
Instructions:
1. Mix space and time until seamless.
2. Fold in matter gently.
3. Add energy; stir vigorously.
4. Bake until life rises.
5. Let cool—consciousness will form on top.
Serve and ponder.
The possibilities of LLMs in education are endless
November 16, 2024 at 1:36 PM
The possibilities of LLMs in education are endless
This sounds difficult to believe for me, based on my empirical negative experience with AI poems - which are effectively always rhyming couples. The article doesn't appear to mention:
1. What prompts were given to the AI
2. How these poems were selected
🤔🤔🧐🧐
1. What prompts were given to the AI
2. How these poems were selected
🤔🤔🧐🧐
November 15, 2024 at 11:17 PM
This sounds difficult to believe for me, based on my empirical negative experience with AI poems - which are effectively always rhyming couples. The article doesn't appear to mention:
1. What prompts were given to the AI
2. How these poems were selected
🤔🤔🧐🧐
1. What prompts were given to the AI
2. How these poems were selected
🤔🤔🧐🧐
Exciting paper from MIT:
1. Typically, LLMs don’t learn during testing, only during training.
2. Researchers used a method called Test-Time Training (TTT), where the model adjusts itself while being tested.
3. With TTT models achieved human performance on the ARC dataset!
Wow!
1. Typically, LLMs don’t learn during testing, only during training.
2. Researchers used a method called Test-Time Training (TTT), where the model adjusts itself while being tested.
3. With TTT models achieved human performance on the ARC dataset!
Wow!
November 12, 2024 at 12:00 PM
Exciting paper from MIT:
1. Typically, LLMs don’t learn during testing, only during training.
2. Researchers used a method called Test-Time Training (TTT), where the model adjusts itself while being tested.
3. With TTT models achieved human performance on the ARC dataset!
Wow!
1. Typically, LLMs don’t learn during testing, only during training.
2. Researchers used a method called Test-Time Training (TTT), where the model adjusts itself while being tested.
3. With TTT models achieved human performance on the ARC dataset!
Wow!
Nvidia introduces Edify Image
High-Quality Image Generation with Pixel Space Laplacian Diffusion Models
huggingface.co/papers/2411....
High-Quality Image Generation with Pixel Space Laplacian Diffusion Models
huggingface.co/papers/2411....
November 12, 2024 at 6:34 AM
Nvidia introduces Edify Image
High-Quality Image Generation with Pixel Space Laplacian Diffusion Models
huggingface.co/papers/2411....
High-Quality Image Generation with Pixel Space Laplacian Diffusion Models
huggingface.co/papers/2411....