Temp / 0.5

exps/trading/baselines.py

@@ -65,6 +65,10 @@ def update_market(config, market):
 def run_exp(task_config, dataset, experiment_name, recorder_name, uri):
 
     # model initiaiton
+    print('')
+    print('[{:}] - [{:}]: {:}'.format(experiment_name, recorder_name, uri))
+    print('dataset={:}'.format(dataset))
+
     model = init_instance_by_config(task_config["model"])
 
     # start exp

lib/layers/positional_embedding.py

@@ -1,11 +1,15 @@
+#####################################################
+# Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.02 #
+#####################################################
 import torch
 import torch.nn as nn
 import math
 
 class PositionalEncoder(nn.Module):
   # Attention Is All You Need: https://arxiv.org/pdf/1706.03762.pdf
+  # https://github.com/pytorch/examples/blob/master/word_language_model/model.py#L65
 
-  def __init__(self, d_model, max_seq_len):
+  def __init__(self, d_model, max_seq_len, dropout=0.1):
     super(PositionalEncoder, self).__init__()
     self.d_model = d_model
     # create constant 'pe' matrix with values dependant on 
@@ -26,4 +30,6 @@ class PositionalEncoder(nn.Module):
   def forward(self, x):
     batch, seq, fdim = x.shape[:3]
     embeddings = self.pe[:, :seq, :fdim]
+    import pdb; pdb.set_trace()
+    outs = self.dropout(x + embeddings)
     return x + embeddings

lib/trade_models/quant_transformer.py

@@ -9,7 +9,6 @@ import numpy as np
 import pandas as pd
 import copy
 from functools import partial
-from sklearn.metrics import roc_auc_score, mean_squared_error
 from typing import Optional
 import logging
 
@@ -23,10 +22,11 @@ from qlib.log import get_module_logger, TimeInspector
 
 import torch
 import torch.nn as nn
+import torch.nn.functional as F
 import torch.optim as optim
 
 import layers as xlayers
-from utils import count_parameters_in_MB
+from utils import count_parameters
 
 from qlib.model.base import Model
 from qlib.data.dataset import DatasetH
@@ -137,9 +137,11 @@ class QuantTransformer(Model):
     mask = ~torch.isnan(label)
 
     if self.loss == "mse":
+      import pdb; pdb.set_trace()
+      print('--')
       return self.mse(pred[mask], label[mask])
-
-    raise ValueError("unknown loss `%s`" % self.loss)
+    else:
+      raise ValueError("unknown loss `{:}`".format(self.loss))
 
   def metric_fn(self, pred, label):
 
@@ -147,8 +149,8 @@ class QuantTransformer(Model):
 
     if self.metric == "" or self.metric == "loss":
       return -self.loss_fn(pred[mask], label[mask])
-
-    raise ValueError("unknown metric `%s`" % self.metric)
+    else:
+      raise ValueError("unknown metric `{:}`".format(self.metric))
 
   def train_epoch(self, x_train, y_train):
 

lib/utils/flop_benchmark.py

@@ -4,10 +4,23 @@ import numpy as np
 
 
 def count_parameters_in_MB(model):
-  if isinstance(model, nn.Module):
-    return np.sum(np.prod(v.size()) for v in model.parameters())/1e6
+  return count_parameters(model, "mb")
+
+
+def count_parameters(model_or_parameters, unit="mb"):
+  if isinstance(model_or_parameters, nn.Module):
+    counts = np.sum(np.prod(v.size()) for v in model_or_parameters.parameters())
   else:
-    return np.sum(np.prod(v.size()) for v in model)/1e6
+    counts = np.sum(np.prod(v.size()) for v in model_or_parameters)
+  if unit.lower() == "mb":
+    counts /= 1e6
+  elif unit.lower() == "kb":
+    counts /= 1e3
+  elif unit.lower() == "gb":
+    counts /= 1e9
+  elif unit is not None:
+    raise ValueError("Unknow unit: {:}".format(unit))
+  return counts
 
 
 def get_model_infos(model, shape):