Hugging Face's logo

01-ai
/
Yi-34B

Text Generation
Transformers
PyTorch
Safetensors
llama
text-generation-inference
Model card Files
xet
 Community
52

run examples for yi-34B-chat, get different results with examples

#41
by cizhenshi - opened
Discussion
cizhenshi

input: hi
response: ## Hello! How can I assist you today? If you have any questions or need information on a specific topic, feel free to ask. ##
which is different with## Hello! How can I assist you today? ## in example

Is this as expected？
YaTharThShaRma999

@cizhenshi it won’t perform super well for a chat model as it’s only a base model. If you want to chat with it, use the yi chat or nous capybara yi34b or dolphin yi 34b
Kai01ai

Hi, @cizhenshi

  1. Check the model name, should be chat model which download from: https://huggingface.co/01-ai/Yi-34B-Chat
  2. Use the chat template(ChatML) may help, you can found example scripts here: https://github.com/01-ai/Yi#31-use-the-chat-model
tianjun changed discussion status to closed
cizhenshi

Hi, @cizhenshi

  1. Check the model name, should be chat model which download from: https://huggingface.co/01-ai/Yi-34B-Chat
  2. Use the chat template(ChatML) may help, you can found example scripts here: https://github.com/01-ai/Yi#31-use-the-chat-model

I use the same code with 2. , but got different result "Hello! How can I assist you today? If you have any questions or need information on a specific topic, feel free to ask."
Kai01ai

Hi, @cizhenshi

  1. Check the model name, should be chat model which download from: https://huggingface.co/01-ai/Yi-34B-Chat
  2. Use the chat template(ChatML) may help, you can found example scripts here: https://github.com/01-ai/Yi#31-use-the-chat-model

I use the same code with 2. , but got different result "Hello! How can I assist you today? If you have any questions or need information on a specific topic, feel free to ask."

This is as expected; the model's responses have a certain degree of uncertainty.

Sign up or log in to comment