GLM-4-32B-0414 - DASH-Q 2-bit GGUF
llama.cpp GGUF files of zai-org/GLM-4-32B-0414 quantized with DASH-Q, at several 2-bit-class sizes. Every file uses only standard llama.cpp tensor types (no tensor above 4 bits) and loads in any recent llama.cpp build.
| File | Type | Size | Bits / weight |
|---|---|---|---|
GLM-4-32B-0414-DASHQ-IQ2_XXS.gguf |
IQ2_XXS | 9.26 GB | 2.27 |
GLM-4-32B-0414-DASHQ-IQ2_XS.gguf |
IQ2_XS | 10.70 GB | 2.63 |
GLM-4-32B-0414-DASHQ-IQ2_M.gguf |
IQ2_M | 11.47 GB | 2.82 |
GLM-4-32B-0414-DASHQ-Q2_K_XL.gguf |
Q2_K_XL | 12.59 GB | 3.09 |
Perplexity (lower is better)
| Type | Model | Size | WikiText-2 | C4 |
|---|---|---|---|---|
| IQ2_XXS | llama.cpp IQ2_XXS (imatrix) | 8.98 GB | 9.97 | 13.78 |
| IQ2_XXS | unsloth UD-IQ2_XXS | 9.26 GB | 10.03 | 13.57 |
| IQ2_XXS | DASH-Q IQ2_XXS | 9.26 GB | 9.06 | 12.99 |
| IQ2_XS | llama.cpp IQ2_XS (imatrix) | 9.90 GB | 9.17 | 12.79 |
| IQ2_XS | DASH-Q IQ2_XS | 10.70 GB | 8.07 | 12.12 |
| IQ2_M | llama.cpp IQ2_M (imatrix) | 11.27 GB | 7.96 | 11.82 |
| IQ2_M | unsloth UD-IQ2_M | 11.47 GB | 7.91 | 11.80 |
| IQ2_M | DASH-Q IQ2_M | 11.47 GB | 7.60 | 11.48 |
| Q2_K_XL | llama.cpp Q2_K (imatrix) | 12.29 GB | 7.70 | 11.65 |
| Q2_K_XL | unsloth UD-Q2_K_XL | 12.81 GB | 7.69 | 11.59 |
| Q2_K_XL | DASH-Q Q2_K_XL | 12.59 GB | 7.61 | 11.36 |
llama-perplexity, context 2048; WikiText-2 test, C4 validation (256 x 2048 tokens).
Usage
llama-cli -m GLM-4-32B-0414-DASHQ-Q2_K_XL.gguf -ngl 99 -c 8192
License
Inherits the license of the base model (zai-org/GLM-4-32B-0414).