Yeming Wen

Ph.D. student at UT CS. Working on code generation.

[Yeming Wen @YemingW]

@jacobaustin132 @polynoamial 4o also can’t do it right. Need reasoning model to get it right because it can calculate multiplication manually.

[Yeming Wen @YemingW]

@AndrewZ45732491 Thanks for the interest! Our code based on Aug 2023's peft & vLLM has compatibility issues due to their huge updates. Still working on the compatibility. For now, you can find FLoRA implementation in our ICLR's supplementary material (peft/src/tuners/flora.py) for reference.

[Yeming Wen @YemingW]

🚀 Excited to share that our paper "Batched Low-Rank Adaptation of Foundation Models (https://t.co/xmqygWBa6V)" was accepted for an oral presentation at #ICLR2024! We're advancing parameter efficient fine-tuning (PEFT) in LLMs for more personalized and efficient AI. (1/n)

[Yeming Wen @YemingW]

🙌 A huge thanks to my advisor @swarat. This work wouldn't have been possible without the joint effort.

[Yeming Wen @YemingW]

🌍 Our experiments on the MultiPL-E code generation benchmark and the multilingual speech recognition tasks demonstrated that FLoRA maintains the accuracy of LoRA at low rank. (7/n)