baidu/ERNIE-4.5-21B-A3B-PT needs about 14.2 GB of memory at Q4_K_M (a 13.2 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 9 quants, from 5.8 GB (IQ2_XXS) to 23.2 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 1.9 GB.
baidu/ERNIE-4.5-21B-A3B-PT is a 21.8B-parameter Ernie model in GGUF format, quantized by ilintar. Its trained context length is 128K tokens. The repository has 9 quant files, from 5.8 GB (IQ2_XXS) to 23.2 GB (Q8_0).
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ2_XXS | 5.8 GB | 6.8 GB | Fits | Get file → | |
| Q2_K_S | 7.4 GB | 8.4 GB | Too big | Get file → | |
| IQ3_M | 9.6 GB | 10.6 GB | Too big | Get file → | |
| Q3_K_M | 10.5 GB | 11.4 GB | Too big | Get file → | |
| IQ4_NL | 12.4 GB | 13.3 GB | Too big | Get file → | |
| Q4_K_M | 13.2 GB | 14.2 GB | Too big | Get file → | |
| Q5_K_M | 15.5 GB | 16.5 GB | Too big | Get file → | |
| Q6_K | 17.9 GB | 18.9 GB | Too big | Get file → | |
| Q8_0 | 23.2 GB | 24.2 GB | Too big | Get file → |
What GPU do you need for baidu/ERNIE-4.5-21B-A3B-PT?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: the best quant that fits is IQ2_XXS (5.8 GB file, about 6.8 GB in use).
- 12 GB VRAM: the best quant that fits is Q3_K_M (10.5 GB file, about 11.4 GB in use).
- 16 GB VRAM: the best quant that fits is Q4_K_M (13.2 GB file, about 14.2 GB in use).
- 24 GB VRAM: the best quant that fits is Q6_K (17.9 GB file, about 18.9 GB in use).
- 32 GB VRAM: the best quant that fits is Q8_0 (23.2 GB file, about 24.2 GB in use).
Which baidu/ERNIE-4.5-21B-A3B-PT quant should you choose?
Q4_K_M is the usual starting point for baidu/ERNIE-4.5-21B-A3B-PT: it balances file size and quality well. Here it is a 13.2 GB file.
The Q8_0 file is 23.2 GB, about 1.8 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is IQ2_XXS at 5.8 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about baidu/ERNIE-4.5-21B-A3B-PT
How much VRAM does baidu/ERNIE-4.5-21B-A3B-PT need?
About 14.2 GB at Q4_K_M with an 8K context window. That is the 13.2 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run baidu/ERNIE-4.5-21B-A3B-PT on an 8 GB GPU?
Yes, with IQ2_XXS: the 5.8 GB file needs about 6.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of baidu/ERNIE-4.5-21B-A3B-PT GGUF?
It depends on the quant: IQ2_XXS is 5.8 GB, Q2_K_S is 7.4 GB, IQ3_M is 9.6 GB, Q3_K_M is 10.5 GB, IQ4_NL is 12.4 GB, Q4_K_M is 13.2 GB and 3 more.
What context length does baidu/ERNIE-4.5-21B-A3B-PT support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.9 GB at 32K.
Where can I download baidu/ERNIE-4.5-21B-A3B-PT GGUF?
The files are on Hugging Face in the ilintar/ERNIE-4.5-21B-A3B-PT-gguf repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does baidu/ERNIE-4.5-21B-A3B-PT use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.