-
Notifications
You must be signed in to change notification settings - Fork 14
[ERROR] Processing message (139318381620712) by ai failed #4
Comments
It seems like you haven't set up Bing with the necessary cookies. |
I added a bing cookie, it shows: signal-aichat-bot | 2023-05-10 18:46:29,212 bot: [INFO] Chat context created for 0n2Ruz0qsQdRFDMlUBP5dyVRhojhmmtgK3HX3U/hXyo=+17c62f31-605f-4fe2-862e-9l10e2a6f27e
signal-aichat-bot | 2023-05-10 18:46:29,465 bot: [INFO] HTTP Request: GET https://edgeservices.bing.com/edgesvc/turing/conversation/create "HTTP/1.1 200 OK"
signal-aichat-bot | 2023-05-10 18:46:29,466 bot: [ERROR] Processing message (139811234150880) by ai failed
signal-aichat-bot | Traceback (most recent call last):
signal-aichat-bot | File "/home/signal-aichat/.local/lib/python3.10/site-packages/semaphore/bot.py", line 128, in _handle_message
signal-aichat-bot | await func(context)
signal-aichat-bot | File "/home/signal-aichat/signal-aichat.py", line 81, in ai
signal-aichat-bot | response = await triggers[default_model](text, ctx)
signal-aichat-bot | KeyError: 'gpt' My .env file: # Signal
SIGNAL_PHONE_NUMBER=+3--------98
# ChatGPT
OPENAI_API_KEY=sk-yPl4---------
# Llama
#CACHE=1
#HOST=0.0.0.0
#PORT=8000
#MODEL=/llama.cpp/models/GPT4xAlpaca.bin
#LLAMA_API_BASE=http://signal-aichat-llama:8000/v1
# Default model options are bing, gpt, or llama
DEFAULT_MODEL=gpt |
The About the Bing cookie, it's not great that currently you have to set it up even if you don't intend to use Bing or else the bot will crash. I'll try to make it possible to configure what models to enable/disable sometime in the coming days. |
That would be cool, I'll be waiting. |
@p6002 This should now be possible with the latest release. |
Thanks, Bing started working and giving answers from internet (with a limit I guess) after couple messages: Sometimes: And maybe can I change Bing language in answered messages? Should be possible i think. GPT quotes and sends a message but only with an error:
|
About Bing API limits, I actually don't know. The GPT error though is a bug that should be fixed now with v0.3.1. |
l'll be waiting, it would also be nice to see the ability to change the language, or the ability to direct the bot with text like in the web version. |
Features like those would have to be implemented in the library I'm using. Though the bot will reply in whatever language your write to it, no? |
Ok I'll take a peek at what you posted, because in addition to gpt bing can also be a useful assistant. |
I managed to install the bot.
It was necessary to change the user in docker-compose from 1000:1000 to 0:0.
I added the phone number, connected via qr code.
In .env I added api openai and default_model to gpt.
When I send a message from the same phone number on the group, the console shows:
signal-aichat-signald | 18:03:25.187 [[redacted 33]9c4-receiver] INFO manager-[redacted 33]9c4 - received sent message transcript from device 1
But no response appears in the chat.
When I text from another phone number, I get a message in the console:
The text was updated successfully, but these errors were encountered: