lazybrickhub

A smaller model starts here.

Explore compression recipes for open models.
Find the right method. Inspect every detail. Make it yours.

All recipes 2

2 recipe versions

Qwen3-4B · AWQ W4A16

v0.1.0

A pinned AWQ W4A16 recipe targeting compressed-tensors export and vLLM inference. Dataset selection remains an explicit planning blocker.

  • Qwen3
  • AWQ
  • W4A16
  • vLLM
  • compressed-tensors
PlannedPlanning blockedNot measured
NVIDIA CUDA · 40 GiB target

Qwen3-0.6B · AWQ W4A16

v0.1.0

A pinned AWQ smoke recipe with explicit calibration and evaluation data, assistant-continuation loss v2, and a declared 16 GiB CUDA target. Execution acceptance remains blocked.

  • Qwen3
  • AWQ
  • W4A16
  • vLLM
  • compressed-tensors
PlannedPlanning blockedNot measured
NVIDIA CUDA · 16 GiB target

From recipe to your stack.

The details you need, before you run.

Get started with lazybrick
  1. 1

    Find a recipe

    Explore models, compression methods, and declared hardware targets.

  2. 2

    Check the details

    Review compatibility, planning blockers, and available evidence.

  3. 3

    Take it with you

    Download the exact YAML and validate it locally with lazybrick.