Phase 5 · Compression

Precision Games

GPTQ → AWQ → FP8 default → NVFP4/MXFP4 frontier.

0/6 · 430 XP

You’re previewing the curriculum. Sign in (top right) to track progress, take knowledge checks, and unlock the verifiers.

Tasks

Read “A Visual Guide to Quantization”
read+50 XPresource ↗
Read GPTQ, AWQ, and SmoothQuant
Hessian-based compensation vs salient-channel scaling vs outlier migration. Interviews ask you to compare them.
paper+100 XPresource ↗
Learn FP8 (E4M3/E5M2) and the 4-bit block formats (NVFP4, MXFP4)
FP8 W8A8 is the boring production default; block-scaled 4-bit is the Blackwell-era shift (GPT-OSS ships in MXFP4).
read+80 XPresource ↗
Read KIVI + the vLLM quantized-KV docs
Why keys quantize per-channel and values per-token. FP8 KV is a near-free 2× cache-capacity win.
paper+60 XPresource ↗
Survey compression beyond quantization: pruning, 2:4 sparsity, distillation
Wanda-style pruning, 2:4 semi-structured sparsity on sparse tensor cores (TRT-LLM ships it), and Minitron-style distillation — which also trains spec-decode drafts.
read+60 XPresource ↗
Pass the quantization drill
Formats, methods, quality tradeoffs, KV quantization. 75% to pass.
drill+80 XPauto-verified