Planned Maintenance
On November 7'th, starting from 1 AM until 3 AM all Mahiru's servers will be down for a planned maintenance and update to version
This will affect both website and the Bot. You won't be able to use any commands or interact with a WebApp during the maintenance period. Any requests to the API will result in 503 status code.
You can always check current status at mahiru.instatus.com.
On November 7'th, starting from 1 AM until 3 AM all Mahiru's servers will be down for a planned maintenance and update to version
1.0.0-b.This will affect both website and the Bot. You won't be able to use any commands or interact with a WebApp during the maintenance period. Any requests to the API will result in 503 status code.
You can always check current status at mahiru.instatus.com.
Update
Maintenance is over, and the new update is here! This is a huge refactor of the GPT module with numerous adjustments, improvements, and fixes.
💬 Meet GPT-4 — Even more powerful
You can now use GPT-4 with Mahiru's /gpt command. It works in the same way as GPT-3 but produces more accurate and detailed outputs. The model can be used in conversation and inline modes. Mahiru does not provide free access to GPT-4, but you can use it with your own API key.
📊 Changes in GPT daily tokens calculations
I have addressed and fixed all issues regarding the calculation of daily token limits. You still have a daily allowance of 20,000 tokens, but with the following changes:
• In conversation mode, each request adds the tokens from all messages to your daily usage. This means that usage increases with the number of messages in a conversation.
• The model-dependent multiplier has been removed. Now, every model consumes the same amount of tokens.
🖥 Improved formatting of the GPT responses
Mahiru is now using Telegram's brand new syntax highlight for code blocks, which allows you to copy any code with just one click. I have also added link embedding, allowing ChatGPT to integrate links into words or sentences.
❌ Old models deprecation
The following OpenAI's text models are now deprecated and can no longer be used with Mahiru's /gpt command:
• text-curie-001
• text-babbage-001
• text-ada-001
• code-davinci-002
• code-cushman-001
📖 Documentation is up to date!
Mahiru's documentation has been updated to match the latest version of the Bot, which you can view as usual at docs.mahiru.one.
That's all for today, see you next time, with even more cool features!👋
1.0.0-bMaintenance is over, and the new update is here! This is a huge refactor of the GPT module with numerous adjustments, improvements, and fixes.
You can now use GPT-4 with Mahiru's /gpt command. It works in the same way as GPT-3 but produces more accurate and detailed outputs. The model can be used in conversation and inline modes. Mahiru does not provide free access to GPT-4, but you can use it with your own API key.
I have addressed and fixed all issues regarding the calculation of daily token limits. You still have a daily allowance of 20,000 tokens, but with the following changes:
• In conversation mode, each request adds the tokens from all messages to your daily usage. This means that usage increases with the number of messages in a conversation.
• The model-dependent multiplier has been removed. Now, every model consumes the same amount of tokens.
Mahiru is now using Telegram's brand new syntax highlight for code blocks, which allows you to copy any code with just one click. I have also added link embedding, allowing ChatGPT to integrate links into words or sentences.
The following OpenAI's text models are now deprecated and can no longer be used with Mahiru's /gpt command:
• text-curie-001
• text-babbage-001
• text-ada-001
• code-davinci-002
• code-cushman-001
Mahiru's documentation has been updated to match the latest version of the Bot, which you can view as usual at docs.mahiru.one.
That's all for today, see you next time, with even more cool features!
Please open Telegram to view this post
VIEW IN TELEGRAM
Unscheduled services reboot
Mahiru is Migrating from v1 to v2 Database. This process requires full reboot of all services. It is going to take 20-30 minutes.
UPD: All services are now up and running.
Mahiru is Migrating from v1 to v2 Database. This process requires full reboot of all services. It is going to take 20-30 minutes.
UPD: All services are now up and running.
Mahiru is down right now. Investigating.
UPD at 3:39 AM: I've identified the issue. Deploying a fix right now. Expecting all services to be back up and running in ~10 minutes.
UPD at 3:47 AM: All services are up. Everything seems to be back to normal🎉
UPD at 3:39 AM: I've identified the issue. Deploying a fix right now. Expecting all services to be back up and running in ~10 minutes.
UPD at 3:47 AM: All services are up. Everything seems to be back to normal
Please open Telegram to view this post
VIEW IN TELEGRAM
Lots of errors on Anime News/Releases server. Investigating.
UPD at 5:30 AM: Doing full restart of all services now.
UPD at 5:45 AM: Seems like I've found the issue. Deploying a fix.
UPD at 5:55 AM: Unfortunately, the fix was unsuccessful. Now I have to shut down everything, until the issue is resolved, in order to prevent any further errors.
UPD at 6.00 AM: All services are now in a maintenance mode. You won't be able to use bot or website until I fix the issue.
UPD at 6:55 PM: I have finally identified the issue. Deploying final fix.
UPD at 7:05 PM: The fix has been deployed. I'm monitoring activity of all services to make sure everything works as intended.
UPD at 7:25 PM: Everything looks fine. The incident has been resolved.