Search results

How to change rank when training glue？

Status: Open.

#38 In yxli2123/LoftQ;

· Rain-yj opened on Aug 20, 2024

When running the train_clm code, there is an explosion in learning rate in the third epoch

Status: Open.

#37 In yxli2123/LoftQ;

· Rain-yj opened on Aug 20, 2024

Identical GPU memory usage and Fine-Tuning time for 2-bit and 4-bit quantized Llama-2-7b-hf model

Status: Open.

#36 In yxli2123/LoftQ;

· zhuzhuxia1221 opened on Aug 17, 2024

Number of iterations seems always set to 1 based on latest code

Status: Open.

#34 In yxli2123/LoftQ;

· au-revoir opened on Jul 4, 2024

About the test result on gsm8k

Status: Open.

#33 In yxli2123/LoftQ;

· lucasliunju opened on Jun 7, 2024

Embedding layer

Status: Open.

#31 In yxli2123/LoftQ;

· ShayekhBinIslam opened on May 26, 2024

Method fails on Gemma-7B model

Status: Open.

#30 In yxli2123/LoftQ;

· Ther-nullptr opened on May 15, 2024

Error with shape

Status: Open.

#27 In yxli2123/LoftQ;

· manlenzzz opened on Apr 21, 2024

Why are base weights on HF LoftQ models in 16-bit?

Status: Open.

#26 In yxli2123/LoftQ;

· RonanKMcGovern opened on Apr 17, 2024

Performance worsens versus QLoRA with TinyLlama

Status: Open.

#25 In yxli2123/LoftQ;

· RonanKMcGovern opened on Apr 15, 2024

Failing to converge when using some random seeds

Status: Open.

#24 In yxli2123/LoftQ;

· Car-pe opened on Apr 13, 2024

Is there any way for using LoftQ to GPTQ or AWQ model?

Status: Open.

#18 In yxli2123/LoftQ;

· WoosungMyung opened on Mar 7, 2024