Van Hans commited on
Commit
8a73165
·
verified ·
1 Parent(s): a60319b

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -5
README.md CHANGED
@@ -16,23 +16,23 @@ The core of the network is a Deep Residual Network (ResNet), a structure well-su
16
 
17
  The network employs a robust torso composed of 20 Residual Blocks. Each block contains convolutional layers with skip connections, allowing for the effective learning of deep, hierarchical features and maintaining stable training.
18
 
19
- ### * Feature Processing:
20
 
21
  The entire network processes data using a high number of channels, specifically 512 filters in its main convolutional layers.
22
 
23
- ### * Input Representation:
24
 
25
  The current board state and history are encoded into a multi-plane tensor with 128 input channels, which typically includes information about piece locations, the player to move, castling rights, and repetition history, a common input format for state-of-the-art chess AI.
26
 
27
- ### * Dual Output Heads:
28
 
29
  The shared ResNet torso branches into two specialized heads:
30
 
31
- ### * Policy Head (p_logits):
32
 
33
  Predicts a probability distribution over the 4672 possible moves (actions) that can be taken. This output is crucial for guiding the Monte Carlo Tree Search (MCTS).
34
 
35
- ### * Value Head (v):
36
 
37
  Outputs a single scalar value, typically between -1.0 (Black is winning) and +1.0 (White is winning). This score represents the network's prediction of the final game outcome from the current position.
38
 
 
16
 
17
  The network employs a robust torso composed of 20 Residual Blocks. Each block contains convolutional layers with skip connections, allowing for the effective learning of deep, hierarchical features and maintaining stable training.
18
 
19
+ * Feature Processing:
20
 
21
  The entire network processes data using a high number of channels, specifically 512 filters in its main convolutional layers.
22
 
23
+ * Input Representation:
24
 
25
  The current board state and history are encoded into a multi-plane tensor with 128 input channels, which typically includes information about piece locations, the player to move, castling rights, and repetition history, a common input format for state-of-the-art chess AI.
26
 
27
+ * Dual Output Heads:
28
 
29
  The shared ResNet torso branches into two specialized heads:
30
 
31
+ * Policy Head (p_logits):
32
 
33
  Predicts a probability distribution over the 4672 possible moves (actions) that can be taken. This output is crucial for guiding the Monte Carlo Tree Search (MCTS).
34
 
35
+ * Value Head (v):
36
 
37
  Outputs a single scalar value, typically between -1.0 (Black is winning) and +1.0 (White is winning). This score represents the network's prediction of the final game outcome from the current position.
38