Quantized GGUF models for UIGEN-T3-4B-Preview
This repository contains GGUF quantized versions of Tesslate/UIGEN-T3-4B-Preview. It uses 128 samples from the training dataset to calibrate iMatrix quants.
Available quantizations:
Full Precision
- FP16 (full precision)
- BF16 (bfloat16 precision)
imatrix Quantizations
These quantizations use importance matrix (imatrix) calibration for better quality:
- IQ3_M-imat
- IQ3_XXS-imat
- Q4_K_M-imat
- Q4_K_S-imat
- IQ4_XS-imat
- Q5_K_M-imat
- Q5_K_S-imat
- Q6_K-imat
- Q8_0-imat
Standard Quantizations
- Q2_K
- Q3_K_L
- Q3_K_M
- Q3_K_S
About imatrix quantization
The imatrix quantizations in this repository use calibration data to preserve the most important weights during quantization, resulting in better model quality compared to standard quantization methods.
Original model
This is a quantized version of Tesslate/UIGEN-T3-4B-Preview.
Generated on
Wed Jun 11 18:08:41 UTC 2025
- Downloads last month
- 141
Hardware compatibility
Log In
to view the estimation
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support