You can deploy this model and start earning money today!

JMingo/
gemma-4-31B-it-qat-GGUF

Name: gemma-4-31B-it-qat-GGUF
Availability: InStock

gguf

gemma4

google/gemma-4-31B-it-qat-q4_0-unquantized

Views

No views yet

Near-lossless GGUF quants of:

Uses the chat template from google/gemma-4-31B-it (2026-05-18), replacing the outdated chat template that was originally bundled with the base QAT model.

Original QAT Specifications

The base QAT model was trained with the following specifications:

Text & Draft Models: 4-bit
Multimodal Projector (mmproj): BF16

Quantization Error Evaluation

Dequantization: Quantized weights were dequantized to F32 for evaluation.
Error Calculation: Quantization error metrics (compared to the BF16 baseline) were computed and accumulated in F64 precision to prevent numerical underflow and precision loss.

Metrics:

MAE (Mean Absolute Error)
RMSE (Root Mean Squared Error)
Max Error

[!NOTE] Interpretation of Metrics: These error metrics do not directly reflect actual model performance. This is because imatrix optimization intentionally increases measured error by downweighting less important parameters to improve overall performance. However, these metrics remain useful for estimating how close the quantized model is to the original, lossless quality.

gemma-4-31B-it-qat

Model	MAE	RMSE	Max Error
`JMingo/gemma-4-31B-it-qat-BF16.gguf` (Baseline)	0.00000000	0.00000000	0.00000000
`JMingo/gemma-4-31B-it-qat-Q4_0.gguf`	0.00000973	0.00001633	0.00061035
`idkwhattoputherenow/gemma-4-31B-it-qat-q4_0-unquantized-q4_0-maxerr.gguf`	0.00001104	0.00001856	0.00062561
`unsloth/gemma-4-31B-it-qat-UD-Q4_K_XL.gguf`	0.00001388	0.00002110	0.00195312
`google/gemma-4-31B_q4_0-it.gguf`	0.00027934	0.00057681	0.01635742
`lmstudio-community/gemma-4-31B-it-QAT-Q4_0.gguf`	0.00027934	0.00057681	0.01635742
`mradermacher/gemma-4-31B-it-qat-q4_0-unquantized.i1-Q4_0.gguf`	0.00001890	0.00006541	0.02943420
`ggnoy/gemma-4-31B-it-qat-q4-bim.gguf`	0.00001890	0.00006589	0.02049255

mtp-gemma-4-31B-it-qat

Model	MAE	RMSE	Max Error
`JMingo/mtp-gemma-4-31B-it-qat-BF16.gguf` (Baseline)	0.00000000	0.00000000	0.00000000
`JMingo/mtp-gemma-4-31B-it-qat-Q4_0.gguf`	0.00002260	0.00003839	0.00128174
`unsloth/mtp-gemma-4-31B-it.gguf`	0.00003191	0.00004994	0.00390625

JMingo/gemma-4-31B-it-qat-GGUF

Original QAT Specifications

Quantization Error Evaluation

gemma-4-31B-it-qat

mtp-gemma-4-31B-it-qat

JMingo/
gemma-4-31B-it-qat-GGUF