422 mistake

[oogxdd]

i'm getting 422 error when trying to fetch from my node server

do you know what might be the reason?

[oogxdd]

im assuming it might be something to do with cross-domain access as on your demo site it works fine

is there a way to play around gpt-j api (from my server) without running my own instance?

[vicgalle]

Hi @maximignatev!

The 422 error seems related to an ill-formed call to the API (https://stackoverflow.com/questions/59929028/python-fastapi-error-422-with-post-request).

How are you calling the API? I provide two alternatives that are working for me, from an external machine:

• Bash:

curl -X 'POST' \
  'http://api.vicgalle.net:5000/generate?context=In%20a%20shocking%20finding%2C%20scientists%20discovered%20a%20herd%20of%20unicorns%20living%20in%20a%20remote%2C%20previously%20unexplored%20valley%2C%20in%20the%20Andes%20Mountains.%20Even%20more%20surprising%20to%20the%20researchers%20was%20the%20fact%20that%20the%20unicorns%20spoke%20perfect%20English.&token_max_length=512&temperature=1&top_p=0.9' \
  -H 'accept: application/json' \
  -d ''

• Python:

import requests

context = "In a shocking finding, scientist discovered a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English."

payload = {
    "context": context,
    "token_max_length": 512,
    "temperature": 1.0,
    "top_p": 0.9,
}
query = requests.post("http://api.vicgalle.net:5000/generate", params=payload)
response = query.json()

print(response)

Let me know if those work for you!

[oogxdd]

Hey. So I been trying to use the api from the node server (next.js api routes)

i'm not sure what was exact reason but i wasn't able to execute the request successfully when using axios library for data fetching but it's working fine now with the default fetch

thanks for reaching out for me

---

that's an offtopic but do you know if i can use advanced parameters like start/stop_sequence and provide examples on how i want model to operate when using your api?

i want to achieve something similar to this: https://github.com/TheProtaganist/gpt-j/blob/main/advanced_usage_template.py

or would you recommend running my own instance and tweaking parameters there?

do you know if i can host it somewhere for free (or with a free trial period at least)?

that must be too many questions 😅 can't stop being curious..

[vicgalle]

Regarding implementing more options with the API, I'm waiting til the model is merged withing huggingface's library (see #1).
But yeah, I'm planing on implementing more features in the near future!

If you want to deploy your own server, you can get TPUs for free by filling this form: https://sites.research.google/trc/. They should answer in a few hours

[oogxdd]

@vicgalle thats awesome thanks!!!