How much VRAM does Tinystories-gpt-0.1-3m need?

Q4_K_M is exactly 8,229,728 bytes (0.01 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

segestic · text

Tinystories-gpt-0.1-3m

segestic/Tinystories-gpt-0.1-3m

Tinystories-gpt-0.1-3m at Q4_K_M is exactly 8,229,728 bytes (0.01 GiB / 0.01 GB) — an effective 17.568 bits per weight, not the nominal 4.

From the file· summed from 1 file(s)From the file· KV per layer

Parameters

Architecture

gpt2

8 layers

Context

—

native (config.json)

License

—

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q2_K	0.01 GiB	7,754,464	16.554	—	afrideva
Q3_K_M	0.01 GiB	7,788,256	16.626	—	afrideva
Q4_K_M	0.01 GiB	8,229,728	17.568	—	afrideva
Q5_K_M	0.01 GiB	8,479,904	18.102	—	afrideva
Q6_K	0.01 GiB	9,544,960	20.376	—	afrideva
Q8_0	0.01 GiB	9,576,704	20.444	—	afrideva

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 0.00 GiB. The real file is 0.01 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json

Layers

Attention heads

—

KV heads

—

Head dim

—

Hidden size

—

Vocab

50,257

Sliding window

none

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does Tinystories-gpt-0.1-3m need?: Q4_K_M is exactly 8,229,728 bytes (0.01 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of Tinystories-gpt-0.1-3m should I use?: Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.