Building blazing-fast async Python tools & APIs for live data pipelines, on-the-fly ML training, crypto-trading engines, apps & Linux automation ππ
#python
How many times in your life have you started something in the terminal using the command `python3`?
Summary of my Python input:
Approximate median = (5 Γ 1 hour) Γ 14 hours Γ (2 years Γ 365 days)
aprox. = 51100 times in range 2 years π«£π·π
The lesson we learned: Not that much computing power is needed. Multiple GPUs wouldn't perform better; ultimately, the system is very resource-efficient, runs smoothly on the CPU, and works perfectly.We also learned that a standard CPU is perfectly adequate for calculating pred..
If you use "a" , "b" from orderbook, pay attention that you dont cheat yourself, your tools, like predictor, with your own fast trade moves, lession learned, all very nice now... Dont know the name for this, but can understand what was happend in my tools....
#TraRyTrade#ADA#ETH#BTC#LSTM#bottrading#LSTM#keras#neverending
IF you got it, dont cp to much and pay attention, before you got a supercomuputer,
Each runner consupt his cpu portion, and may if you cp and start more, other are touched in speed and performance...!! ;-)
#TraRyTrade#keras
Fighting with model stability,but think may found resolution. That is the last part to fix.... may somebody can feel with me on that thematik (epochs opt SGD use_ema Dropout(x.x)x Adam or ...swish relu merged) base build or reuse model, fit .learning_rate