"TinyZero has demonstrated its ability to perform multiplication tasks"
So it's already better than ChatGPT.
So it's already better than ChatGPT.
February 1, 2025 at 1:20 PM
"TinyZero has demonstrated its ability to perform multiplication tasks"
So it's already better than ChatGPT.
So it's already better than ChatGPT.
UC Berkeley Scientists Replicate DeepSeek AI for Just $30 interestingengineering.com/innovation/u...
DeepSeek replicated for peanuts: US scientists achieve feat in $30
TinyZero achieves impressive results with minimal resources, raising questions about the cost of AI development.
interestingengineering.com
February 1, 2025 at 3:58 AM
UC Berkeley Scientists Replicate DeepSeek AI for Just $30 interestingengineering.com/innovation/u...
UC Berkeley did it.
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
GitHub - Jiayi-Pan/TinyZero: Clean, minimal, accessible reproduction of DeepSeek R1-Zero
Clean, minimal, accessible reproduction of DeepSeek R1-Zero - Jiayi-Pan/TinyZero
github.com
February 1, 2025 at 5:11 PM
UC Berkeley did it.
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
Reinforcement Learning is on the rise in 2025. Rewarding steps along the way, not just a final output, is unlocking new "reasoning" capabilities.
Researchers at @ucberkeleyofficial.bsky.social have released #TinyZero, reproducing #DeepSeek R1-Zero:
Researchers at @ucberkeleyofficial.bsky.social have released #TinyZero, reproducing #DeepSeek R1-Zero:
GitHub - Jiayi-Pan/TinyZero: Clean, accessible reproduction of DeepSeek R1-Zero
Clean, accessible reproduction of DeepSeek R1-Zero - Jiayi-Pan/TinyZero
buff.ly
January 31, 2025 at 4:42 PM
Reinforcement Learning is on the rise in 2025. Rewarding steps along the way, not just a final output, is unlocking new "reasoning" capabilities.
Researchers at @ucberkeleyofficial.bsky.social have released #TinyZero, reproducing #DeepSeek R1-Zero:
Researchers at @ucberkeleyofficial.bsky.social have released #TinyZero, reproducing #DeepSeek R1-Zero:
Phd studerende fra Berkeley genskaber DeepSeeks “aha moment” med en 3B LLM og 30$ compute. Det der reasoning LLM halløj er måske i virkeligheden slet ikke så mega svært 👌😎 Tjek koden her: github.com/Jiayi-Pan/Ti... #dkai #dktech
GitHub - Jiayi-Pan/TinyZero: Clean, minimal, accessible reproduction of DeepSeek R1-Zero
Clean, minimal, accessible reproduction of DeepSeek R1-Zero - Jiayi-Pan/TinyZero
github.com
February 6, 2025 at 5:52 PM
Phd studerende fra Berkeley genskaber DeepSeeks “aha moment” med en 3B LLM og 30$ compute. Det der reasoning LLM halløj er måske i virkeligheden slet ikke så mega svært 👌😎 Tjek koden her: github.com/Jiayi-Pan/Ti... #dkai #dktech
Side note: we saw some nice adoption from DeepSeek-R1 reprod repos, which is humbling, if not thrilling!
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
GitHub - Jiayi-Pan/TinyZero: Clean, minimal, accessible reproduction of DeepSeek R1-Zero
Clean, minimal, accessible reproduction of DeepSeek R1-Zero - Jiayi-Pan/TinyZero
github.com
February 5, 2025 at 5:02 PM
Side note: we saw some nice adoption from DeepSeek-R1 reprod repos, which is humbling, if not thrilling!
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
Some Berkeley researchers managed to replicate DeepSeek for $30. I bet tomorrow some dudes in the Ukraine with a Chromebook and an old PS3 still capable of running Linux is going to figure out how to get Claude results consuming power equivalent of a TV remote push.
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
GitHub - Jiayi-Pan/TinyZero: Clean, accessible reproduction of DeepSeek R1-Zero
Clean, accessible reproduction of DeepSeek R1-Zero - Jiayi-Pan/TinyZero
github.com
January 29, 2025 at 5:30 PM
Some Berkeley researchers managed to replicate DeepSeek for $30. I bet tomorrow some dudes in the Ukraine with a Chromebook and an old PS3 still capable of running Linux is going to figure out how to get Claude results consuming power equivalent of a TV remote push.
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
Por esto ha sido tan disruptora la aparición de DeepSeek.
El día de mañana, cuando las grandes tecnológicas te vendan su mierda con letras de neón, no será porque no tienes alternativas libres.
Dos tips o mejor nombres a extraer del artículo:
1- TinyZero
2- Qwen
El día de mañana, cuando las grandes tecnológicas te vendan su mierda con letras de neón, no será porque no tienes alternativas libres.
Dos tips o mejor nombres a extraer del artículo:
1- TinyZero
2- Qwen
Consiguen recrear todo el potencial de DeepSeek. Les ha costado menos de 30 euros
Si no te has pasado la última semana viviendo debajo de una roca, en un retiro en un monasterio budista en el Tíbet sin acceso a un móvil o a Internet, probabl
www.lavanguardia.com
February 15, 2025 at 10:27 PM
Por esto ha sido tan disruptora la aparición de DeepSeek.
El día de mañana, cuando las grandes tecnológicas te vendan su mierda con letras de neón, no será porque no tienes alternativas libres.
Dos tips o mejor nombres a extraer del artículo:
1- TinyZero
2- Qwen
El día de mañana, cuando las grandes tecnológicas te vendan su mierda con letras de neón, no será porque no tienes alternativas libres.
Dos tips o mejor nombres a extraer del artículo:
1- TinyZero
2- Qwen
TinyZero – Researchers Replicated DeepSeek’s R1-Zero Model for Just $30
TinyZero - Researchers Replicated DeepSeek’s R1-Zero Model for Just $30
cybersecuritynews.com
February 5, 2025 at 8:44 AM
TinyZero – Researchers Replicated DeepSeek’s R1-Zero Model for Just $30
TinyZero: reproduction of DeepSeek R1-Zero for countdown and multiplication tasks using reinforcement learning
June 26, 2025 at 6:34 PM
TinyZero: reproduction of DeepSeek R1-Zero for countdown and multiplication tasks using reinforcement learning
Nel mondo dell'intelligenza artificiale, le novità continuano a susseguirsia un ritmo vertiginoso, e TinyZero è una delle innovazioni più#apprendimentoautomatico #apprendimentoperrinforzo #democratizzazioneIA #github #IA #IAlowcost #intelligenzaartificialeeconomica
www.corrierenerd.it/tinyzero/
www.corrierenerd.it/tinyzero/
February 4, 2025 at 7:32 AM
Nel mondo dell'intelligenza artificiale, le novità continuano a susseguirsia un ritmo vertiginoso, e TinyZero è una delle innovazioni più#apprendimentoautomatico #apprendimentoperrinforzo #democratizzazioneIA #github #IA #IAlowcost #intelligenzaartificialeeconomica
www.corrierenerd.it/tinyzero/
www.corrierenerd.it/tinyzero/
Using AI to make decisions too I'd guess... github.com/markoelez/Ti...
GitHub - markoelez/TinyZero: Clean, minimal, accessible reproduction of DeepSeek R1-Zero
Clean, minimal, accessible reproduction of DeepSeek R1-Zero - markoelez/TinyZero
github.com
February 4, 2025 at 4:01 PM
Using AI to make decisions too I'd guess... github.com/markoelez/Ti...
La boîte de Pandore a été ouverte par deepseek 😅 Jiayi pan vient de sortir Tinyzero une méthode capable de re-entraîner des modèles 3B ou 0.5B à la manière de Deepseek zéro. Ça donne un modèle sympa pour moins de 30$ 😯 #ia #deepseek #mistral #llm
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
GitHub - Jiayi-Pan/TinyZero: Clean, accessible reproduction of DeepSeek R1-Zero
Clean, accessible reproduction of DeepSeek R1-Zero - Jiayi-Pan/TinyZero
github.com
January 31, 2025 at 6:58 AM
La boîte de Pandore a été ouverte par deepseek 😅 Jiayi pan vient de sortir Tinyzero une méthode capable de re-entraîner des modèles 3B ou 0.5B à la manière de Deepseek zéro. Ça donne un modèle sympa pour moins de 30$ 😯 #ia #deepseek #mistral #llm
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
This is pretty incredible, they reproduced DeepSeek R1-Zero in the CountDown game for < $30 and the results seem to scale with model size/quality. Does this represent an order of magnitude jump in performance/size?
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...
GitHub - Jiayi-Pan/TinyZero
Contribute to Jiayi-Pan/TinyZero development by creating an account on GitHub.
github.com
January 24, 2025 at 8:28 PM
This is pretty incredible, they reproduced DeepSeek R1-Zero in the CountDown game for < $30 and the results seem to scale with model size/quality. Does this represent an order of magnitude jump in performance/size?
github.com/Jiayi-Pan/Ti...
github.com/Jiayi-Pan/Ti...