Troubleshooting

First aid kit to help you quickly resolve issues

The fact that you reached this page probably means that you encountered issues with PROMPTMETHEUS software, let's try to resolve quickly!

Select

  • Problems with FORGE
  • Problems with ARCHERY
  • How to report problems

Problems with the FORGE playground

First of all, please go through the following checklist, which will resolve most of the common issues.

Quick fix checklist

  1. Check the system status to make sure that the FORGE service is operational.
  2. Make sure you have entered your OpenAI API key correctly.
  3. Make sure that your API key has access to the model you have selected. You might need to get on a waitlist to get access for GPT-4 for instance, or for models with large context windows.
  4. Make sure that you do not have a VPN activated that might interfere with requests.

System checks

If the quick fixes didn't do it, please also run the following checks and report the results as specified below.

  1. If you have, try to open the app in a different browser.
  2. Try to create and execute a very simple prompt, e.g. "Tell me a joke." If that does work, there is probably an issue with your prompt. Are you using any special characters? Try to remove them and see if that helps. If it does, please let us know so that we can address the issue.
  3. Open the app in an incognito browser window and try to execute the example prompt. This will make sure that all data is reset and no browser extensions interfere with the app. If that works, there is most likely an issue with corrupted data that the browser is holding on to. Please get in touch so that I can help you out with this.

Common pitfalls

My data is gone

FORGE stores all data in your browser. That means, if you open the app in a different browser or on a different device your data will not be available. In order to use your prompts, datasets, etc. on multiple devices, you can import/export your data via the respective actions in the account menu or switch to a Standard plan on ARCHERY, which automatically syncs data between devices.

Known issues

There are some issues with input duplication on Firefox. We're working on it, but for a quick fix, please try a different browser.

Maintenance mode

FORGE is currently in maintenance mode, meaning it will not receive any new features and only a limited amount of bug fixing. To get the latest features and improvements, please switch to a Standard plan on ARCHERY.

Problems with the ARCHERY cloud platform

First of all, please go through the following checklist, which will resolve most of the common issues.

Quick fix checklist

  1. Make sure you have a stable internet connection, you can test it here.
  2. Refresh the page.
  3. Sign out and back in.
  4. Check the system status to make sure that the ARCHERY service is operational.
  5. Make sure you have entered the LLM provider API keys correctly.
  6. Make sure that your API key has access to the model you have selected. You might need to get on a waitlist to get access for GPT-4 for instance, or for models with large context windows.
  7. Make sure that you do not have a VPN activated that might interfere with requests.

System checks

If the quick fixes didn't do it, please also run the following checks and report the results as specified below.

  1. If you have, try to open the app in a different browser.
  2. Try to create and execute a very simple prompt, e.g. "Tell me a joke." If that does work, there is probably an issue with your prompt. Are you using any special characters? Try to remove them and see if that helps. If it does, please let us know so that we can address the issue.
  3. Open the app in an incognito browser window and try to execute the example prompt. This will make sure that no browser extensions interfere with the app.

Known issues

There have been instances where Zscaler security software interferes with the app and causes problems. If you use Zscaler in your network and cannot log in, please get in touch.

How to report problems

The best way to report problems is via email or in the Discord or Slack communities. It helps a lot if you can attach a screenshot of the issue and ideally check if the browser reports any errors in the developer console. Please have a look that this nice guide to find out how that works in your browser.

Please also include the following information in your report, it will help a lot with the investigation:

  • What device are you working on? Desktop, laptop, tablet, mobile, etc.
  • What operating system are you using? Windows, OS X, Linux, etc. If it is not a recent version, please also include the version.
  • Which browser are you using?
  • If there is anything special about your setup that might be related to the problem, please let me know.
  • If you have experienced similar issues with other software, please also let me know.

If you find any vulnerabilities in our systems, please check the security page and disclose them confidentially at security@promptmetheus.com.

Supported LLMs

Anthropic

Claude 2.1

Claude 2

Claude Instant 1.2

Claude Instant 1

Cohere

Command / Nightly

Command Light

OpenAI

GPT-4 Turbo

GPT-4 / 32k

GPT-3.5 Turbo / 16k

GPT-3.5 Turbo Instruct

DaVinci 003

Curie 001

Babbage 001

Ada 001

Perplexity

Llama 2 70B

Code Llama 34B

Mistral 7B Instruct

OpenHermes 2.5 Mistral 7B

OpenHermes 2 Mistral 7B

pplx 70B chat alpha

pplx 7B chat alpha

PaLM 2

Text Bison

Chat Bison

Code Bison

NLP Cloud

Chat Dolphin

Dolphin

Aleph Alpha

Luminous Supreme

Luminous Extended

Luminous Base

AI21 Labs

Jurassic 2 Ultra

Jurassic 2 Mid

Jurassic 2 Light

Deep Infra

Llama 2 70B Chat HF

Llama 2 13B Chat HF

Llama 2 7B Chat HF

Mistral 7B Instruct v0.1

xAI

coming soon...

Hugging Face

coming soon...

Replicate

coming soon...

Azure

coming soon...

Bedrock

coming soon...

Custom

coming soon...

Local

coming soon...