UltraChat

UltraChat copied to clipboard

Hi team, Thank you for the great work. I'm wondering if you can post the prompt to interact with user's question during data collection. I saw the system prompt for...

lingchensanwen

Questions about system prompt

2

I observe that the inference script contains an embedded system prompt. Is it also included in training? I wonder how much the performance will be affected if changing the system...

ningshaoling

greate job, Tks you so much for sharing, I have a question about prompt engineering that You make a dialog around 1 topic like you say: "We sample 200k specific...

trangtv57

Hello! Thanks for the awesome work! I am trying to find the metadata, e.g. topic, corresponding to each data point. They don't seem to be present in the current released...

luyug

大佬生成数据集的脚本能不能开放一下，怎么用gpt3对话生成的代码

3

929359291

Human prompts for dataset generation.

3

> We instruct the user model with carefully designed prompts to mimic human user behavior @ningding97 could you provide these prompts you used? Thanks!

winglian

你好，非常感激开源训练数据。 我想请教下你们训练13b模型使用的超参是多少？训练了多少步数？

wjx-git

运行报错

2

从hugging face上下载了模型 运行chat_cli.sh 模型回复出现乱码问题 这是为什么？ 

bjtuxck

def get_model_tokenizer(args): model = LlamaForCausalLM.from_pretrained(args.model_name_or_path) tokenizer = LlamaTokenizer.from_pretrained(args.model_name_or_path) tokenizer.add_special_tokens({'pad_token': ""}) model.resize_token_embeddings(len(tokenizer)) model = bmt.BMTrainModelWrapper(model) return model, tokenizer 假设在单机8卡服务器上，加载UltraChat65B的模型进行微调，会不会存在OOM的问题？每个卡都会执行model = LlamaForCausalLM.from_pretrained(args.model_name_or_path)加载一份模型，哪怕存CPU内存，65B大概需要130G的内存，8卡差不多需要1T的内存，而服务器总内存也差不多1T。

jinmin527

getting syntax error from weight_diff.py

1

sh-4.2$ sudo sh recover.sh File "weight_diff.py", line 23 special_tokens_dict: Dict, ^ SyntaxError: invalid syntax

Sumit5194