This is pretty neat. How complex can the tasks be? I'm also impressed by the model's capability in the demo considering it's only a 3B model. Curious to see what a larger local model like gpt-oss-20b would be able to accomplish, hosted via a local API on, say, Ollama.
This looks cool and I am also impressed with the tiny Qwen models.
I wonder if there is a malicious use of this similar to how people did bitcoin mining in visitors browsers. Maybe someone will create a botnet of compute slaves to run 470B models in a distributed way ;)
This is pretty neat. How complex can the tasks be? I'm also impressed by the model's capability in the demo considering it's only a 3B model. Curious to see what a larger local model like gpt-oss-20b would be able to accomplish, hosted via a local API on, say, Ollama.
This looks cool and I am also impressed with the tiny Qwen models.
I wonder if there is a malicious use of this similar to how people did bitcoin mining in visitors browsers. Maybe someone will create a botnet of compute slaves to run 470B models in a distributed way ;)