Search results

Embedding int8: -375MB, cosine sim 0.9999

Status: Open.

#190 In Vahe1994/AQLM;

· thedorohindorohin-ctrl opened on May 23, 2026

30+ compression experiments on AQLM codes — all failed. One thing worked.

Status: Open.

#189 In Vahe1994/AQLM;

· thedorohindorohin-ctrl opened on May 23, 2026

Add devstral small 2 (mistral3) support

Status: Open.

#185 In Vahe1994/AQLM;

· lexasub opened on Feb 15, 2026

How to get vector quantized in output_dim?

Status: Open.

#184 In Vahe1994/AQLM;

· shanhx2000 opened on Oct 20, 2025

Compatibility with Qwen models

Status: Open.

#183 In Vahe1994/AQLM;

· AIpplier opened on Oct 16, 2025

Release new version as Intel XPU supported

Status: Open.

#182 In Vahe1994/AQLM;

· jiqing-feng opened on Aug 20, 2025

AQLM for VLM

Status: Open.

#181 In Vahe1994/AQLM;

· LugerW-A opened on Jul 18, 2025

Only Use AQ part

Status: Open.

#178 In Vahe1994/AQLM;

· zzzzzzzyb opened on Jun 1, 2025

he calculation of the compression ratio

stale

Status: Open.

#177 In Vahe1994/AQLM;

· maoshanwen opened on Apr 30, 2025

Is two-bit quantization only achievable under specific conditions?

stale

Status: Open.

#174 In Vahe1994/AQLM;

· maoshanwen opened on Apr 19, 2025

Quantized model reverts to original size after converted to HF format

stale

Status: Open.

#173 In Vahe1994/AQLM;

· QianChef opened on Apr 19, 2025