Alan May

I'm a little confused, `warmup_ratio = 0.03` or `warmup_ratio = 0.04`? According to train_vicuna_13b.sh : https://github.com/lm-sys/FastChat/blob/a26db3c814889035d92c8ae80d6defbd7381ee55/scripts/train_vicuna_13b.sh#L17 But the README use: `warmup_ratio 0.03`. Also train-vicuna.yaml: https://github.com/lm-sys/FastChat/blob/a26db3c814889035d92c8ae80d6defbd7381ee55/scripts/train-vicuna.yaml#L111 Thanks!

Fix #130

Finetune command: ```shell cd scripts composer train/train.py \ ./train/yamls/finetune/mpt-7b_dolly_sft.yaml ``` Got error: ```txt Traceback (most recent call last): ... .../llm-foundry/llmfoundry/data/finetuning/dataloader.py", line 215, in _validate_config raise ValueError( ValueError: In the dataset...

Feature request：support ExLlama

7

ExLlama (https://github.com/turboderp/exllama) It's currently the fastest and most memory-efficient executor of models that I'm aware of. Is there an interest from the maintainers in adding this support?

[Bug] gen with regex: Token fusion between input and output, try to avoid this by removing the space at the end of the input.

3

### Checklist - [X] 1. I have searched related issues but cannot get the expected help. - [X] 2. The bug has not been fixed in the latest version. -...