Quiz · 4 questions

🔄 Conversion & Quantization

Hugging Face → MLX, plus the quantization choices that matter

Level 0Curious

0 XP0/51 lessons0/15 achievements

0/100 XP to next level100 XP to go0% complete

Quiz

01Which file in an MLX-format model directory is the source of truth for the model's architecture and quantization config?

Hint

It's the file mlx-lm reads to dispatch to the right model class.

02Which mlx_lm.convert flag combination produces a 4-bit quantized model with group size 64 and uploads it to your HF repo?

Hint

All four pieces — quantize on, bits, group, upload — need to be present.

03Lesson 3 names Q4 with group_size 64 as the sane default for 7B+ instruct models on Apple Silicon. When should you graduate to a higher precision (Q8 or full)?

Hint

It's a measurement-driven decision, not a default-driven one.

04When is GGUF the right format choice over MLX format?

Hint

Format choice follows deployment target.

← PreviousMLX-format vs GGUF — When to Pick Which 🎯 Fine-Tuning — LoRA, QLoRA, DoRA, and Shipping the Result →LoRA — Why You Don't Train the Whole Model Anymore

Spotted a bug or have feedback on this page?Report an Issue

Comments 0

🔔 Reply notifications (sign in)

No comments yet — be the first.