dsaunders23 commited on
Commit
9558ddf
·
verified ·
1 Parent(s): e8ea682

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +7 -4
README.md CHANGED
@@ -13,14 +13,14 @@ tags:
13
 
14
  # Chess Position Evaluation Dataset
15
 
16
- This dataset contains chess positions in FEN format along with their evaluations and expected outcomes. It is formatted in the Alpaca instruction-following format.
17
 
18
  ## Dataset Structure
19
 
20
  Each example contains:
21
  - instruction: A prompt to evaluate the chess position
22
  - input: The FEN string representing the chess position
23
- - output: The evaluation value and expected winner
24
 
25
  ## Example
26
 
@@ -28,10 +28,13 @@ Each example contains:
28
  {
29
  "instruction": "Evaluate the following chess position in FEN format.",
30
  "input": "rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - 0 1",
31
- "output": "0.52 (Black wins)"
32
  }
33
  ```
34
 
35
  ## Intended Use
36
 
37
- This dataset can be used to train models to evaluate chess positions and predict game outcomes.
 
 
 
 
13
 
14
  # Chess Position Evaluation Dataset
15
 
16
+ This dataset contains chess positions in FEN format along with their Stockfish evaluations. It is formatted in the Alpaca instruction-following format.
17
 
18
  ## Dataset Structure
19
 
20
  Each example contains:
21
  - instruction: A prompt to evaluate the chess position
22
  - input: The FEN string representing the chess position
23
+ - output: The evaluation score in pawn units (positive values favor White, negative values favor Black)
24
 
25
  ## Example
26
 
 
28
  {
29
  "instruction": "Evaluate the following chess position in FEN format.",
30
  "input": "rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - 0 1",
31
+ "output": "0.52"
32
  }
33
  ```
34
 
35
  ## Intended Use
36
 
37
+ This dataset can be used to train models to evaluate chess positions. The evaluation scores are in pawn units, where:
38
+ - Positive values indicate an advantage for White
39
+ - Negative values indicate an advantage for Black
40
+ - Values closer to 0 indicate a more balanced position