Text Generation
Transformers
PyTorch
English
llama
Inference Endpoints
text-generation-inference

specific instruct prompt to use

#2
by digitous - opened

Alpaca is well known to use

Instruction:

Response:

And a lot of people say, with WizardLM and Vicuna it's
Human:
Assistant:

However, looking at WizardLM's own training data they link to from their Git, it's
human:
gpt:

Can you help show the specific instruct prompt that works with Manticore and Mega?
Thank you for your AWESOME work!

Check the model card. Same as llama:

### Instruction: <prompt>
### Assistant:
Open Access AI Collective org

Manticore gets fed several different prompts, so it's pretty good at figuring out whichever one you use.

Ah very cool. I was curious if several instruct types blended together would yield a model that has an intrinsic understanding of a generalized instruct format. Thank you for your hard work!

winglian changed discussion status to closed

Sign up or log in to comment