My-Knowledge-Base

My notes and reflections on LLMs, agents, and related topics.

Agentic RL

MoE RL

Quantization

• INT4 QAT RL Training: Introduces the INT4 QAT RL end-to-end practice, including the technical details and the implementation of the INT4 QAT RL end-to-end practice.
• Activation-aware Weight Quantization: Introduces AWQ, which reduces low-bit quantization error by identifying activation-important channels and applying channel-wise scaling before quantization.

Slime

• CPU Usage: Analyzes the cpu usage in Qwen3-30B-A3B MoE training.

Evaluation

• GAIA 2: Introduces the basic understanding of GAIA2.

Scheduling and Routing

• Speculative Decoding In RL