FastChat

FastChat copied to clipboard

ZYHowell

accroding to data_cleaning.md, how to get sharegpt_20230322_html.json firstly?

1

eeric

How long does it take for the llama team to respond to the weight request form?

3

Does anyone get the pre-trained weight of the model, I submitted the form almost 1 week and still don't get any response?

nguenvantuan

`encoding` parameter in `logging.basicConfig` breaks Python 3.8

3

In #303, `encoding='utf-8'` is passed to `logging.basicConfig`. This change breaks Python 3.8 because `encoding` parameter is not available until Python 3.9. Reference: https://docs.python.org/3/library/logging.html#logging.basicConfig

icyblade

accroding to data_cleaning.md, after perform split_long_conversation.py, what do we do next?

2

eeric

Missing import from train/train.py for LORA training

1

The train_lora.py import a function `smart_tokenizer_and_embedding_resize` that was deleted in previous commit.

jeremycochoy

I modified the conversation.py file to fix the accidental removal of the logic for the v0 weights. Using v0 stopped functioning correctly after the introduction of v1.1. Now both versions...

laidybug

Can I run this on a 7900xtx and 5700xt which have 24gb and 8gb of vram respectively? I kept my old gpu in my computer when I upgraded.

sirus20x6

When we apply delta, we load two complete models at the same time, which puts a lot of strain on the CPU memory. This PR allows us to apply delta...

andy-yang-1

The conversion command cannot locate the installed transformers

2

Hi, when run the conversion command for 7B to generate the diff between llama and vicuna, it gives error as below from transformers import AutoTokenizer, AutoModelForCausalLM ModuleNotFoundError: No module named...

ant3001