3 points | by 0in 3 hours ago ago
2 comments
cant they just use open source deepseek if it already benches better than open ai lol i dont get
1. Everyone has been doing distillation for a long time
2. You ideally want outputs from multiple models, not a single one
3. Distillation (or a model trace) is insufficient on its own (a) you need a sufficiently strong base (b) crafting RL rewards is an art
4. You are conflating DeepSeek with Moonshot (K3)
cant they just use open source deepseek if it already benches better than open ai lol i dont get
1. Everyone has been doing distillation for a long time
2. You ideally want outputs from multiple models, not a single one
3. Distillation (or a model trace) is insufficient on its own (a) you need a sufficiently strong base (b) crafting RL rewards is an art
4. You are conflating DeepSeek with Moonshot (K3)