What GPU do I need to run shibing624/macbert4csc-base-chinese?

102M parameters, published in F32. View on Hugging Face

102M
Parameters
F32
Native precision
BertForMaskedLM
Architecture
text-generation
Pipeline

macbert4csc-base-chinese is published by shibing624 on Hugging Face, with 541,994 downloads and 120 likes to date. It's a BertForMaskedLM model built for text-generation, published natively in F32.

VRAM required & cheapest live GPU fit

Required VRAM = weight size at each precision times a flat 1.2 overhead for activations and fragmentation. The KV-cache grows with context and is not in that factor; it is listed per model below where the architecture is published. Full formula and assumptions: methodology.

PrecisionWeight sizeRequired VRAMCheapest live fitGPUs neededEst. $/hr (full fit)
FP320.4 GB0.5 GBRTX 30701$0.088/hr
FP8 (quantized)0.1 GB0.1 GBRTX 40701$0.154/hr
INT4 (quantized)0.0 GB0.1 GBRTX 30701$0.088/hr

A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.

INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.

Cheapest way to run macbert4csc-base-chinese at its published (F32) precision: 1× RTX 3070, at $0.088/hr per GPU ($0.088/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.

macbert4csc-base-chinese: KV cache by context length

The KV cache is the memory the attention layers hold for every token of context, on top of the weights. It is not part of the flat 1.2x overhead in the table above, grows with context length and with every concurrent request, and is why a long-context deployment needs more VRAM than the table shows.

KV-cache size for macbert4csc-base-chinese is not published for this architecture, so no figure is shown.

macbert4csc-base-chinese: common questions

How much VRAM does macbert4csc-base-chinese need?

0.5 GB at FP32, 0.1 GB at FP8 (quantized), 0.1 GB at INT4 (quantized). All of those sit under 6 GB, the smallest capacity this page reasons about, so VRAM is not what limits where this model runs. The 0.4 GB of weights plus inference overhead is the whole requirement.

Can macbert4csc-base-chinese run in 16-bit instead of FP32?

Yes. Its published weights are FP32, 0.4 GB, or 0.5 GB once inference overhead is added. Loading the same weights in 16 bits halves that to 0.2 GB, or 0.2 GB with overhead. How much accuracy the cast costs is model-specific and is not measured here.

How many copies of macbert4csc-base-chinese fit on one RTX 3070?

17, by VRAM alone. That card carries 8.0 GB and one copy needs 0.5 GB at FP32, on a live rate of $0.088/hr for the whole card. Throughput is not modelled here, so 17 copies is not 17 times the requests served.

Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.

shibing624 models

Alternatives at this size

Other models for text-generation within about a third of macbert4csc-base-chinese's 102M parameters, from other model lines.

More on macbert4csc-base-chinese

Related reading: RTX 3070 pricing and specs, How much VRAM you need for LLMs, Serving LLMs with vLLM, vLLM vs TensorRT-LLM vs SGLang, and Best GPU for LLM inference.

Your own GPU cloud accounts

One console for the GPU clouds you already rent from. Your keys stay on your machine.

$20 a month, flat

Connect your accounts

One console for the GPU clouds you already use.

Connect your accounts for $20 a month, flat. Your provider bills the GPU.

© 2026 Aquanode. All rights reserved.

All trademarks, logos and brand names are the property of their respective owners.