Search results

[Bug]:Llama 3.1 8B Instruct W2A16 Quantization leads to Model Collapse (repeating "!") despite using "Best" settings

bugSomething isn't workingSomething isn't working

Status: Open.

#1342 In intel/auto-round;

· manfeilong opened on Jan 26, 2026

[Feature]: Support Re‑Quantizing Per‑Tensor FP8 Models to Other Dtypes

good first issueGood for newcomersGood for newcomers

Status: Open.

#1337 In intel/auto-round;

· yiliu30 opened on Jan 25, 2026

·  

1.0.0

[Enhancement]: Refactor the NVFP4 weight global scale calculation to be block-wise

enhancementNew feature or requestNew feature or request

Status: Open.

#1331 In intel/auto-round;

· yiliu30 opened on Jan 23, 2026

·  

1.0.0

[Feature]: support hadamard transform when calibration & tuning

enhancementNew feature or requestNew feature or request

Status: Open.

#1323 In intel/auto-round;

· lkk12014402 opened on Jan 23, 2026

Resolve CUDA UT failure with transformers 5.0.0rc

Status: Open.

#1315 In intel/auto-round;

· xin3he opened on Jan 22, 2026

·  

0.10.0

Resolve CPU UT failure with transformers 5.0.0rc

high priority

Status: Open.

#1314 In intel/auto-round;

· xin3he opened on Jan 22, 2026

·  

0.10.0

Ensuring correct device dispatch after module replacement

good first issueGood for newcomersGood for newcomers

Status: Open.

#1308 In intel/auto-round;

· yiliu30 opened on Jan 21, 2026

Add support for re-quantize MXFP4/MXFP8 models

Status: Open.

Feature

#1306 In intel/auto-round;

· yiliu30 opened on Jan 21, 2026

·  

0.10.0

Preparations for New Model

Status: Open.

#1305 In intel/auto-round;

· yiliu30 opened on Jan 21, 2026

The source package contains shared libraries and the wheel naming is incorrect

Status: Open.

#1296 In intel/auto-round;

· dhuangnm opened on Jan 16, 2026

[feat] For FP8 models, keep layers specified in <code>ignore_layers</code> in their original FP8 format

good first issueGood for newcomersGood for newcomers

Status: Open.

Feature

#1284 In intel/auto-round;

· yiliu30 opened on Jan 15, 2026

<code>ignore_layers</code> does not take effect on FP8 model

bugSomething isn't workingSomething isn't working

Status: Open.

Bug

#1283 In intel/auto-round;

· yiliu30 opened on Jan 15, 2026

·  

1.0.0