Release 2.0.1

Hi everyone,

As those of you that follow our socials know, this April will be a very important time for us and the subnet!! (https://x.com/zeussubnet/status/2031769355210612945?s=20). Zeus is launching its first pilot with a European energy trading firm! For that, there are three changes we need to make.

• Extending the forecast period to 15 days into the future
• Adding another variable Surface Solar Radiation Downwards, SSRD with weight of 0.2. This variable is measured in W/m², so make sure you start training your models for the correct units!
• Giving priority to Europe (i.e. artificially increasing the errors in Europe by 1.5)

Release 2.0.0

Hi everyone,

Thank you for participating in the testnet phase over the past week. We have fixed the reported bugs, optimized performance, and officially pushed Zeus v2.0.0 to Mainnet.

🛠️ Release Highlights
Simplified Setup: Miners and validators no longer need an Open-Meteo API key to operate.
Compression Updates: We have refined the Synapse compression to improve data syncing reliability.
System Fairness: The mandatory hash-commitment system is fully active to ensure fairness and eliminate relay mining.
⚖️ Scoring Weights
Our "Winner Takes All" model continues with the following variable weights:
2m_temperature: 0.2
100m_u_component_of_wind: 0.4
100m_v_component_of_wind: 0.4

📌 Action Items:
Update your nodes: Please run git pull to get the latest commit.
Check the docs: Refer to the updated README for the exact schedule of the 4x-daily, 48-hour forecasting requests.

Thank you for your continued support and feedback!

Happy mining!!

Release 1.5.6

🛠️ Hotfix 1.5.6 - Validator Error Patch 🛠️

Just a quick hotfix today! An unused import in `setup.py` was causing some annoying errors for our validators. We've pushed a patch to remove it and stabilize things. 

A thank you to the **Yuma team** for catching this and bringing it to our attention.

**Live Now in 1.5.6:**
* **Validator Fix:** Removed the problematic unused import causing validator errors. No other changes were made to the codebase.

**Action for Miners:**
* **No action required.** This does not impact you at all. You can ignore this update and keep using the burn period to optimize your setups purely for accuracy! 

**Action for Validators:**
* **Auto-update:** Your nodes should update automatically.
* **Manual update:** Pull the latest code and restart to resolve the import error. 

The Zeus team

Release 1.5.5

We are initiating a temporary 90% incentive burn. This is step one of a massive architectural shift to make Zeus a true enterprise-grade forecasting oracle (specifically for energy trading).

Live Now in 1.5.5:

• 90% Incentive Burn: Emissions are temporarily slashed to protect the network's economics while we transition the architecture.

Coming Soon (Why we are doing this):

• True Future Forecasting: We are killing the "-5 to 0 days" ERA5 delay meta. The new focus will be strictly on predicting the future.
• Speed is Decoupled: Response time will no longer impact your score. The API will be unlinked from live responses, meaning latency becomes irrelevant. Accuracy is everything.

Action for Miners:
Do not panic. Your current ML models are NOT obsolete. In upcoming updates, only the forecast format and evaluation window will change. Use this burn period to optimize your setups purely for accuracy instead of speed.

Action for Validators:

• Auto-update: Your nodes will update automatically.
• Manual update: Pull the latest code and restart.

This is a painful but necessary transition to bring Zeus to the next level and secure high-value B2B clients. The transition starts now.

The Zeus team

Release 1.5.4

⚡Release 1.5.4 - Incentivizing future challenges more⚡

For our downstream products, challenges further into the future are the most interesting. However, these challenges are naturally more difficult, and miners attempting them shouldn't be penalised compared to very short term forecasts. Previously we already incorporated challenge age into the gamma curving shift to account for this, but we have now improved this mechanism.

Additionally, we have increased the importance of the temperature variable, since it's crucial for our potential energy trader clients. Wind has been decreased slightly since both wind variables are highly correlated, so both temperature and wind are now equally important.

Action for Miners:
Nothing! Happy mining ⛏️

The Zeus team

Release 1.5.3

⚡Release 1.5.3 - Setting EMA alpha to normal value after previous temporary resolve⚡

Previously we had to temporarily increase the minimal moving average alpha, to allow miners to more quickly recover their normal performance after having been unfairly punished due to OpenMeteo's/ERA5's downtime. This was always a temporary resolution, and this minimal alpha value has now been reset to its proper default. Please see the notebook for an in-depth explanation of what this parameter does.

Action for Miners:
Nothing! Happy mining ⛏️

The Zeus team

Release 1.5.2

⚡Release 1.5.2 - Resolve issues regarding OpenMeteo's downtime⚡

Today, for the first time in the history of the subnet, OpenMeteo's api malfunctioned. It didn't crash, but secretly returned NaN values, resulting in challenges that incorrectly yielded a zero quality score for all miners. This update introduces additional safeguards, ensuring that in the future such challenges are filtered out. Additionally, we temporarily increase the moving average alpha, so that miners can more quickly recover their previous incentive.

Action for Miners:
Nothing! Happy mining ⛏️

The Zeus team

Release 1.5.1

⚡Release 1.5.1 - ⏳ Far future challenges are scored more fairly⚡

This is a rather small update, with no changes required for miners. Through careful analysis, we became aware that miners who competed in the subnet for a longer time period could get penalised too much by challenges that are more into the future. These challenges are naturally more difficult, and due to the nature of the subnet only miners who have been in the subnet for a while could possibly be scored for these challenges. To remedy this, we are using the gamma difficulty scaling, which previously represented geographical difficulty, to also reflect temporal difficulty. This will make it so challenges which are further into the future are scored more fairly compared to short-window challenges.

Action for Miners:
Nothing! Happy mining ⛏️

The Zeus team

Release 1.5.0

⚡Release 1.5.0 - 🌡️ New surface pressure variable incoming! ⚡
This update introduces a new variable, the surface pressure (in Pascal) at the surface of the earth! We might potentially implement some other small changes, if so we will announce them here too. The update will go live sometime this week, giving our miners some time to update.

Action for Miners:
Update your miner code to also predict the surface pressure, depending on the variable in the synapse. Note that the base-miner code will already work, but keep in mind that you can likely improve on this! 🚀

The Zeus team

Release 1.4.3

⚡Release 1.4.3 - Focus slightly less on speed and small stability improvements⚡

After analysing the current performances on our subnet and discussing with our miners, we have decided to slightly reduce the importance of the response speed.

What changes?

• Responsive speed now makes up 20% of the incentive, with the remaining 80% focussed on RMSE
• A response speed of 0.4 seconds of less is now considered a perfect score for this component
• The wind speed baseline has improved slighty, also simplying the respective OpenMeteo code

For validators
The codebase should now be compatible with a Torch-based metagraph, although running a Numpy based metagraph is still preferred.

The zeus team