faunix qwen38 27b distillation 40k
quantization aware healing
qwen3 8 max glm5 2 kimi k3 distillation
thinking machines inkling small
qwen38 max distillation 50k
jackrong deepseek v4 distill 8000x
perplexity agent training
needle 26m distilled gemini
semantic self distillation for language model uncertainty
rethinking on policy distillation one training example
ai4ai test time capability transfer
codifying the judge
y combinator garry tan wants us open weight ai labs to distill frontier models