Model troubleshooting

Home Website Guide


To search in the rentry with keywords (on computer):

  • On Windows: CTRL + F
  • On Mac: Command + F


Any model

Model's maximum context size

The problem: "Error during generation, error: This model's maximum context length is is 4097 tokens. However, your messages resulted in 4100 tokens. Please reduce the length of your message", with any number.

The solution: You set your context size over your model's maximum context size. Lower your context size.

  1. Press the three lines on the top right corner of your chat window.
  2. Select 'Configuration'.
  3. Select 'Generation parameters'.
  4. Set your context size to a value smaller than your model's maximum context size.
  5. Save.

Model max context size

Character definition too large

The problem: "Error during generation, error: Error: Character definition is too large".

The solution: The definition of your character is too large for your model.

  • Trim down the character definition.
  • Check if your context size is set at the right value (no, it shouldn't be set to 0). It means not smaller than the bot's size, and not bigger than your model's size. Seriously, check it. It might have been set to 2k tokens.
  • It seems to be an issue with example dialogs, they are not discarded when the context is completely used. If necessary, trim down the example dialogs.

400

The problem: "400" is generated instead of text.

The explanation: Happens with some proxies. The model takes too long to generate text, longer than Chub's limit. No solution if you want to keep using it. If you are willing to try something else, Chub has its own free model.

Mars & Mercury

No credits nor subscription

The problem: "Error during generation, error: Error: You have neither trial credits nor an active subscription for this model"

The solution: You are trying to use Mars or Mercury without having a subscription nor credits. Subscribe, or change your API. If you already have a subscription, check if you are subscribed to Mercury. If you are subscribed to Mercury, check if you selected Mercury, since you don't have access to Mars. For reference:
Mercury = $5
Mars = $20

Reading 'ID'

The problem: "Error during generation, error: TypeError: Cannot read properties of undefined (reading 'id')"

The solution: You are trying to make the bot generate a message without any chat history. Send a message or refresh the page.

Max new tokens 0

The problem: When setting the max new tokens to 0, it changes back to a full number.

The solution: Chub's models don't handle 0 for unlimited. You need to have a full number.

Failed to fetch

The problem; When using a Chub API on Janitor AI, as a reverse proxy, you get "Failed to fetch" or "Network error".

The solution: Add /chat/completions at the end of your proxy link.

Character definition too large/model's maximum context size with reverse proxy

The problem: When using a Chub API on Janitor AI as a reverse proxy, you get an error with token numbers or definition size.

The solution: Issue on Janitor AI's side. They have their context size set to 9k and you can't edit it, while Chub's models are 8k tokens.

Malformed tokens

The problem: "Error during generation, error: Error: Error during inference. Server may be overloaded, or your input too large or malformed".

The solution: Could be a server issue, check the status. Or, it could be an issue with the model selection. Try changing your model, saving, and changing back to the one you want to use.

Syntax error / Unexpected token

The problem: "Error during generation, error: SyntaxError: Unexpected token '<', "<html><bod>",... is not valide JSON".

The solution: Issue in your preset. Use one of the recommended ones instead.

Anthropic

Invalid credentials

The problem: "Error during generation, error: Error: Invalid credentials"

The solution: Use a new key.

Overloaded

The problem: "overloaded_error: Overloaded"

The solution: Issue on Anthropic's side. Check their status page.

OpenAI

The model does not exist

The problem: You encounter a message such as "Error during generation, error: Error: The model 'claude-v1' does not exist" or "Error during generation, error: Error: The model 'mobile' does not exist".

The solution: It means your model hasn't been set up correctly.

  1. Press the three lines on the top right corner of your chat window.
  2. Select 'Configuration'.
  3. Press 'Prompt structure'.
  4. Check if the right API is selected in the 'API' dropdown menu. (OpenAI should be selected)
  5. Select a model in the 'Model' drop down menu. (One of OpenAI's models should be selected)
  6. Save.

Model doesn't exist

Validation error

The problem: When trying to generate a reply, your get this: "Error during generation, error: Error: 3 validation errors for Request body stop str type expected (type=type_error.str) body stop ensure this value has at least 1 items (type=value_error.list.min_items; limit_value=1) body -> stop ensure this value has at least 1 items (type=value_error.list.min_items; limit_value=1)".

The solution: You are trying to generate with GPT 4 Vision, which is a model made for pictures. Change your model.

Can't generate

The problem: "Empty response from API"

The solution: You hit the filter. Change your jailbreak.

Invalid credentials

The problem: "Error during generation, error: Error: Invalid credentials"

The solution: Use a new key.

Quota exceeded

The problem: "You exceeded your current quota, please check your plan and billing details."

The solution: You don't have any credit left on your OpenAI account. Add credits or change models.

OpenAI doesn't give free trials anymore, even for new accounts.

Empty response

The problem: "Empty response received from API."

The solution: It is very likely you hit the filter. Get a better jailbreak.

Kobold

Rep pen

The problem: "{"rep_pen":["Must be greater than or equal to 1."]}.

The solution: Set your repetition penalty to 1 or more. Kobold can't handle less than that.

Invalid URL

The problem: "Error during generation, error: TypeError: Failed to construct 'URL': Invalid URL"

The solution: You need to add an URL in Secrets.

Ooba

Reading 'name'

The problem: "Error during generation: TypeError: Cannot read properties of undefined (reading 'name')"

The solution: If you have something in your author's notes or in the character's notes, erase it.

OpenRouter

Reading'0'

The problem: "Error during generation: TypeError: Cannot read properties of undefined (reading 'content')" or "Error during generation: TypeError: Cannot read properties of undefined (reading '0')"

The solution: It typically indicates that you are attempting to use a paid model without sufficient funds in your account, or your account balance is in the negative. Add funds to your account.

Reading'content'

The problem: "Error during generation: TypeError: Cannot read properties of undefined (reading 'content')"

The solution:

  • It may indicates that you are attempting to use a paid model without sufficient funds in your account, or your account balance is in the negative. Add funds to your account.
  • It may indicate you are hitting the filter of your model. Either change your jailbreak or change your model.
  • It may indicate your model doesn't work. Try changing the model.

My model default to Mistral

The problem: When trying to generate something, Mistral generates the message instead of the model I want.

The solution: You are using a filtered model, and you hit the filter. Mistral is the fallback model, which is automatically used in this case. Either use a better jailbreak or change your model.

Rate limit exceeded

The problem: "Error during generation, error: Error: Rate limit exceeded: free-models-per-day."

The solution: Free OpenRouter models have a limit of 10 requests per minutes, 100 requests per day.

  • Wait a few minutes.
  • Wait for a day.
  • Use a paid model (no limit).

Authentification header

The problem: "Error during generation, error: Error: Missing Authentication header or invalid API."

The solution: Under "Secrets", no reserve proxy option should be selected. (Both OpenAI and Anthropic need to be on "My own API key")

Google

Filtered by Google

The problem: "Error during generation, error: Error: Response filtered by Google for reason: OTHER".

The solution: You hit the filter, you need a better jailbreak.

NovelAI

Readable stream is locked

The problem: "Error during generation, error: TypeError: Failed to execute 'getReader' on 'ReadableStream': ReadableStreamDefaultReader constructor can only accept readable streams that are not yet locked to a reader".

The solution:

  • Your max new tokens is set to 0. Set it to a full number.
  • You selected the wrong model.
  • If it doesn't work, try using a new key.

Body stream already read

The problem: "Error during generation, error: TypeError failed to execute 'json' on 'Response': body stream already read."

The solution: In your generation parameters, you have your minimum length set to 0. Leave it blank. Or, you context size is wrong.

Short messages

The problem: No matter what is the max new tokens, messages are always short.

The solution: It's normal, NovelAI can only generate 150 tokens at a time. You can press the "Continue" button.

Edit
Pub: 07 Jun 2024 22:06 UTC
Edit: 01 Dec 2024 11:28 UTC
Views: 2987