ShiYu
ShiYu
I have also encountered this problem, and the current solution is: first convert the elements of type datetime64[ns] in bins to INT through the pd.Timestamp.value method; after obtained the result...
The cut-off date for the training set is February 2024 (inclusive). The validation set spans from March to May 2024 (inclusive), and the out-of-sample period begins in June 2024.
Thank you, some artificial intelligence conferences are target conferences, such as IJCAI, NeurIPS, AAAI, etc.
Similar to transformer based tokenizer, but using causal CNN instead of transformer block
just apply 1D convolution on the OHLCVA. input (B, T, E) -> (B, E, T) -> causal CNN -> (B, C, T) -> linear -> (B, T) -> tokenize ->...
Yes, this is right.
我不清楚你这边微调的具体细节。但是如果采样温度T较低，或者sample count比较大，那理论上mean values就是会更加平滑
这段K线单边下跌，直接预测的话上涨的概率肯定大于继续下跌，所以模型肯定是预测上涨。毕竟只有这么一段单边下跌的K线信息，让人判断也会觉得上涨的概率更大。
都可以，建议使用conda
Many thanks for reaching out and for your interest in our work. Regarding your question about the dataset, unfortunately, due to our licensing agreements with financial data vendors, we are...