Comment by Reason077
2 months ago
> "Anti-distillation: injecting fake tools to poison copycats"
Plot twist: Chinese competitors end up developing real, useful versions of Claude's fake tools.
2 months ago
> "Anti-distillation: injecting fake tools to poison copycats"
Plot twist: Chinese competitors end up developing real, useful versions of Claude's fake tools.
I cannot bring myself to care about distillation, when these companies have built their empires on top of everyone else's stolen data, while at the same time telling the world they're out to replace us all.
Sure, AI progress comes to a halt then as everyone switches to the copycats that can't innovate, and the frontier companies are bled dry.
"frontier" as in the frontier of using everybody else's code, books, art of everyone else for a specific purpose that was never intended to, as in, not even open source projects ever imagined LLMs becoming a thing and their licenses reflected as much.
2 replies →
These companies don't get the chance to raise a trillion dollars, and you're laughing???
Poor babies.
Again, I don’t care about them.
11 replies →
Qwen are the only guys doing real innovation. (LLM architectures and such.)
Everyone else is just gaming engagement metrics and benchmarks.
Amazing that people on HN can't distinguish between training a model on open source data vs distilling a model's outputs.
You must have some absolutely unhinged ideas about what "open source" means.
Tbh, I think distillation is happening both ways. And at this stage, "quality" is stagnating, the main edge is the tooling. The harness of CC seems to be the best so far, and I wonder if this leak would equalize the usability.
This was my favorite bit, "We're going to steal countless copy righted works and completely ignore software licenc... wait, what? You aren't allowed to turn around and do it to us! Stop that right now!"
‘You can’t fight in here. This is the War Room!’
Has Claude stopped claiming to be deepseek when prompted in Chinese yet? It wasn't long that it hit the news and blogs
Definitely. We can expect zAI, Qwen, Minimax CCs very soon
more likely, they would parse them out using simple regex, the whole point is they're there but not used. Distillation is becoming less common now however