DEV Community

Post-training Digest Series' Articles

Back to Felipe 0liveira's Series
GRPO leaves the chatbot and starts fixing OCR, while DPO and GRPO get debugged from the inside
felipe0liveira profile

GRPO leaves the chatbot and starts fixing OCR, while DPO and GRPO get debugged from the inside

#ai #deeplearning #llm #machinelearning
Comments
6 min read
Distillation's Hidden Costs and Payoffs — Teachers Leak Behavior, Reward Models Surpass Them
felipe0liveira profile

Distillation's Hidden Costs and Payoffs — Teachers Leak Behavior, Reward Models Surpass Them

#ai #llm #machinelearning
1
Comments 1
5 min read
GRPO keeps winning the RL-vs-data argument, and LoRA gets a mechanistic audit
felipe0liveira profile

GRPO keeps winning the RL-vs-data argument, and LoRA gets a mechanistic audit

#ai #deeplearning #llm #machinelearning
1
Comments
7 min read
Teaching GRPO to Doubt Itself — Calibrated Fixes for Noisy Graders and Reward Hacking
felipe0liveira profile

Teaching GRPO to Doubt Itself — Calibrated Fixes for Noisy Graders and Reward Hacking

1
Comments
6 min read
On-policy distillation is quietly eating GRPO's lunch — and DPO's tax bill just got itemized
felipe0liveira profile

On-policy distillation is quietly eating GRPO's lunch — and DPO's tax bill just got itemized

Comments
8 min read