What's Changed

• Meta MobileLLM model support added.
• lm-eval[gptqmodel] integration merged upstream.
• Intel/IPEX cpu inference merged replacing QBits (deprecated).
• Auto-fix/patch ChatGLM-3/GLM-4 compat with latest transformers.
• New .load() and .save() api. 100% model CI coverage and unit testing.

Note that 1.2.1 and 1.2.0 are the same. 1.2.0 release had a bad version name: 1.2.1-dev embedded in code/release.

PR Logs

• [KERNEL] [CPU] Replace QBits with IPEX by @jiqing-feng in #527
• [MODEL] add mobilellm support by @LRL-ModelCloud in #509
• [MODEL] [FIX] chatglm-3 and glm-4 compat with latest transformer by @LRL-ModelCloud in #555
• [API] Replace from_pretrained and from_quantized with unified load() by @ZYC-ModelCloud in #535
• [FIX] throw exception when avg_loss is NaN by @LRL-ModelCloud in #514
• Remove exllama in gptqmodel_ext folder by @ZYC-ModelCloud in #521
• Save quant log to csv by @CSY-ModelCloud in #524
• [CI] Use lm-eval for model regression tests by @PZS-ModelCloud in #518

New Contributors

• @jiqing-feng made their first contribution in #527

Full Changelog: v1.1.0...v1.2.1