baidu · image

ERNIE-Image-Turbo

baidu/ERNIE-Image-Turbo

ERNIE-Image-Turbo at Q4_K_M is exactly 5,019,124,416 bytes (4.67 GiB / 5.02 GB) — an effective 4.998 bits per weight, not the nominal 4.

From the file· summed from 1 file(s)
Parameters
8.0B
Architecture
wan
Context
native (config.json)
License
apache-2.0

Shipped quantizations

exact bytes, summed from published files
QuantSizeExact bytesEffective bpwTensorsPublisher
Q2_K2.96 GiB3,178,545,8563.165unsloth
Q3_K_S3.57 GiB3,834,135,2323.818unsloth
Q3_K_M3.64 GiB3,909,632,7043.893unsloth
UD-Q2_K3.71 GiB3,979,002,5603.962unsloth
Q4_04.44 GiB4,762,747,5844.743unsloth
UD-Q3_K_M4.47 GiB4,799,218,3684.779unsloth
Q4_K_S4.54 GiB4,875,993,7924.856unsloth
Q4_K_M4.67 GiB5,019,124,4164.998unsloth
Q4_14.89 GiB5,254,267,5845.232unsloth
Q5_05.35 GiB5,745,787,5845.722unsloth
UD-Q4_K_M5.38 GiB5,778,817,7285.755unsloth
Q5_K_S5.46 GiB5,861,655,2325.837unsloth
Q5_K_M5.53 GiB5,932,958,4005.908unsloth
Q5_15.81 GiB6,237,307,5846.211unsloth
UD-Q5_K_M6.27 GiB6,736,560,8326.708unsloth
Q6_K6.32 GiB6,790,267,5846.762unsloth
Q8_08.10 GiB8,694,907,5848.659unsloth
F1614.96 GiB16,067,707,58416.001unsloth
BF1614.96 GiB16,067,707,58416.001unsloth

Pipeline components

a diffusion model is a graph of parts, not one file
ComponentSizeShareCan live on the CPU?
text encoder7.17 GiB32%yes — the usual first VRAM saving
vae0.16 GiB1%no, must be resident
denoiser14.96 GiB67%no, must be resident
Full pipeline22.29 GiBresident if nothing is offloaded

The parameter count published for a diffusion model describes the denoiser alone. Running it also requires its text encoder and VAE, and the text encoder is often nearly as large as the denoiser — which is why offloading it is the standard first move when you run out of memory.

We publish component sizes here, not throughput. Community-submitted image-generation rates do exist for many GPUs and we show them on the hardware pages, but they aggregate runs at different resolutions, step counts and settings, so they cannot be attributed to one model. Peak memory during sampling is unmeasured by any public source, and we do not estimate it.

Compare with

same modality, comparable size

Will it run on your card?

full quant x context sweep

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 4.21 GiB. The real file is 4.67 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

Architecture unavailable — this repository is gated and no ungated mirror was found. Exact file sizes above are still authoritative; only the KV math needs the config.

Questions people ask

How much VRAM does ERNIE-Image-Turbo need?
Q4_K_M is exactly 5,019,124,416 bytes (4.67 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of ERNIE-Image-Turbo 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.