File size: 13,340 Bytes
7f9c451 12471a6 2c0ecb8 12471a6 38d6b68 2c0ecb8 7f9c451 2c0ecb8 12471a6 7f9c451 12471a6 7f9c451 12471a6 38d6b68 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 12471a6 2c0ecb8 12471a6 2c0ecb8 7f9c451 38d6b68 7f9c451 2c0ecb8 7f9c451 12471a6 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 38d6b68 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 2c0ecb8 7f9c451 38d6b68 12471a6 7f9c451 2c0ecb8 38d6b68 2c0ecb8 12471a6 7f9c451 2c0ecb8 7f9c451 2c0ecb8 12471a6 38d6b68 7f9c451 2c0ecb8 38d6b68 12471a6 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 | from __future__ import annotations
import chess
import gradio as gr
from gradio_chessboard import Chessboard
START_FEN = chess.STARTING_FEN
CENTER4 = [chess.D4, chess.E4, chess.D5, chess.E5]
PIECE_V = {
chess.PAWN: 1.0,
chess.KNIGHT: 3.2,
chess.BISHOP: 3.3,
chess.ROOK: 5.0,
chess.QUEEN: 9.0,
chess.KING: 0.0,
}
# HDS-ish "axes" (kept human-readable)
AXES = [
"material",
"king_safety",
"piece_activity",
"center_reach",
"initiative",
"tactical_tension",
]
W = {
"material": 1.8,
"king_safety": 1.4,
"piece_activity": 1.0,
"center_reach": 1.0,
"initiative": 0.9,
"tactical_tension": 1.1,
}
# -----------------------------
# Helpers
# -----------------------------
def clamp(x: float, lo: float = -1.0, hi: float = 1.0) -> float:
return max(lo, min(hi, float(x)))
def side_name(color: bool) -> str:
return "White" if color == chess.WHITE else "Black"
def square(sq: int) -> str:
return chess.square_name(sq)
def uci_to_label(prev: chess.Board, mv: chess.Move) -> str:
s = f"{square(mv.from_square)}→{square(mv.to_square)}"
if prev.is_capture(mv):
s += " (capture)"
if prev.is_castling(mv):
s += " (castle)"
return s
def infer_move(prev_fen: str, new_fen: str) -> chess.Move | None:
b = chess.Board(prev_fen)
for mv in b.legal_moves:
bb = b.copy(stack=False)
bb.push(mv)
if bb.fen() == new_fen:
return mv
return None
def count_attackers(board: chess.Board, color: bool, sq: int) -> int:
return len(board.attackers(color, sq))
# -----------------------------
# Axis components (deterministic, engine-less)
# -----------------------------
def _material(board: chess.Board, me: bool) -> float:
s = 0.0
for pt, v in PIECE_V.items():
s += v * (len(board.pieces(pt, me)) - len(board.pieces(pt, not me)))
return s # raw
def _king_exposure(board: chess.Board, me: bool) -> float:
# Higher risk => worse
ksq = board.king(me)
if ksq is None:
return 10.0
opp = not me
risk = 0
# king neighborhood attacked?
for sq2 in chess.SquareSet(chess.BB_KING_ATTACKS[ksq]):
if board.is_attacked_by(opp, sq2):
risk += 1
return float(risk) # 0..8-ish
def _castled(board: chess.Board, me: bool) -> bool:
ksq = board.king(me)
if ksq is None:
return False
return (ksq in [chess.G1, chess.C1]) if me == chess.WHITE else (ksq in [chess.G8, chess.C8])
def _minor_development(board: chess.Board, me: bool) -> int:
# "awake pieces" (minors moved off initial squares)
if me == chess.WHITE:
kn0 = {chess.B1, chess.G1}
bi0 = {chess.C1, chess.F1}
else:
kn0 = {chess.B8, chess.G8}
bi0 = {chess.C8, chess.F8}
k = sum(1 for sq in board.pieces(chess.KNIGHT, me) if sq not in kn0)
b = sum(1 for sq in board.pieces(chess.BISHOP, me) if sq not in bi0)
return k + b # 0..4
def _center_reach(board: chess.Board, me: bool) -> int:
# number of center squares attacked by me (0..4)
c = 0
for sq2 in CENTER4:
if board.is_attacked_by(me, sq2):
c += 1
return c
def _mobility(board: chess.Board, me: bool) -> int:
b = board.copy(stack=False)
b.turn = me
return b.legal_moves.count()
def _initiative(board: chess.Board, me: bool) -> float:
# crude: checks + captures ratio for side-to-move
b = board.copy(stack=False)
b.turn = me
total = 0
checks = 0
caps = 0
for mv in b.legal_moves:
total += 1
if b.is_capture(mv):
caps += 1
bb = b.copy(stack=False)
bb.push(mv)
if bb.is_check():
checks += 1
if total == 0:
return 0.0
return 0.65 * (checks / total) + 0.35 * (caps / total)
def _tactical_tension(board: chess.Board, me: bool) -> float:
# crude: number of captures available now + hanging targets
b = board.copy(stack=False)
b.turn = me
caps = 0
for mv in b.legal_moves:
if b.is_capture(mv):
caps += 1
# scale
return float(min(10, caps))
def eval_axes(board: chess.Board, me: bool) -> dict[str, float]:
mat = _material(board, me) # raw
kex = _king_exposure(board, me) # raw (lower is better)
dev = _minor_development(board, me) # 0..4
cen = _center_reach(board, me) # 0..4
mob = _mobility(board, me) # ~0..60
ini = _initiative(board, me) # 0..1
ten = _tactical_tension(board, me) # 0..10
# normalize each axis to [-1, 1] (deterministic, stable)
axes = {
"material": clamp(mat / 10.0), # ±10 pawns range
"king_safety": clamp((4.0 - kex) / 4.0), # kex 0..8 => safety roughly
"piece_activity": clamp((dev - 1.5) / 1.5), # dev 0..4
"center_reach": clamp((cen - 1.5) / 1.5), # cen 0..4
"initiative": clamp((ini - 0.12) / 0.25), # ini 0..1
"tactical_tension": clamp((ten - 2.0) / 4.0), # ten 0..10
}
return axes
def score(axes: dict[str, float]) -> float:
return sum(W[a] * axes.get(a, 0.0) for a in AXES)
def top_deltas(before: dict[str, float], after: dict[str, float], k: int = 3):
rows = []
for a in AXES:
d = after.get(a, 0.0) - before.get(a, 0.0)
wd = d * W[a]
rows.append((a, d, wd))
rows.sort(key=lambda x: abs(x[2]), reverse=True)
return rows[:k], rows
def axis_pretty(a: str) -> str:
return {
"material": "Material",
"king_safety": "King safety",
"piece_activity": "Piece activity",
"center_reach": "Center reach",
"initiative": "Initiative",
"tactical_tension": "Tactical tension",
}.get(a, a)
def axis_table_md(before: dict[str, float], after: dict[str, float]) -> str:
lines = []
lines.append("| Axis | Before | After | Δ |")
lines.append("|---|---:|---:|---:|")
for a in AXES:
b = before.get(a, 0.0)
c = after.get(a, 0.0)
d = c - b
lines.append(f"| {axis_pretty(a)} | {b:+.2f} | {c:+.2f} | {d:+.2f} |")
return "\n".join(lines)
# -----------------------------
# Risk / explanation (HDS-ish causal trace)
# -----------------------------
def risk_flags(prev: chess.Board, mv: chess.Move, after: chess.Board) -> list[str]:
flags = []
mover = prev.turn
opp = not mover
# moved piece vulnerability (very simple)
p = after.piece_at(mv.to_square)
if p and p.color == mover:
atk = count_attackers(after, opp, mv.to_square)
dfd = count_attackers(after, mover, mv.to_square)
if atk > dfd:
flags.append("Moved piece becomes a likely target (more attacked than defended).")
# early queen move warning
p0 = prev.piece_at(mv.from_square)
if p0 and p0.piece_type == chess.QUEEN and prev.fullmove_number <= 6:
flags.append("Early queen move: can invite tempo hits.")
# king safety regression
if _king_exposure(after, mover) > _king_exposure(prev, mover):
flags.append("King safety slightly worsens (more pressure near your king).")
return flags
def advantage_label(axes: dict[str, float], me: bool) -> str:
# label from "me" perspective using combined score
s = score(axes)
if s >= 0.55:
return "Ahead"
if s >= 0.20:
return "Slightly better"
if s > -0.20:
return "Roughly equal"
if s > -0.55:
return "Slightly worse"
return "Behind"
def short_verdict(before_axes: dict[str, float], after_axes: dict[str, float], me: bool) -> str:
b = advantage_label(before_axes, me)
a = advantage_label(after_axes, me)
# 1-liner: state transition
if b == a:
return f"State: {a} (no major swing)."
return f"State shift: {b} → {a}."
def causal_sentence(top3, flags: list[str]) -> str:
# Build a "because / at the cost of" line using top weighted deltas
pos = [t for t in top3 if t[2] > 0]
neg = [t for t in top3 if t[2] < 0]
because = ", ".join(axis_pretty(x[0]) for x in pos[:2]) if pos else "structure"
cost = ", ".join(axis_pretty(x[0]) for x in neg[:2]) if neg else ""
if cost:
core = f"Because {because} improves, at the cost of {cost}."
else:
core = f"Because {because} improves with limited downside."
if flags:
# add the first risk as a hard boundary
core += f" Boundary risk: {flags[0]}"
return core
# -----------------------------
# Candidate moves (Top-3)
# -----------------------------
def candidate_score(prev: chess.Board, mv: chess.Move) -> tuple[float, dict[str, float], list[str]]:
me = prev.turn
before_axes = eval_axes(prev, me)
b2 = prev.copy(stack=False)
b2.push(mv)
after_axes = eval_axes(b2, me)
base = score(after_axes)
# small deterministic bonuses
bonus = 0.0
if prev.is_capture(mv):
bonus += 0.06
if b2.is_check():
bonus += 0.06
if prev.is_castling(mv):
bonus += 0.05
# small deterministic risk penalty
flags = risk_flags(prev, mv, b2)
penalty = 0.04 * len(flags)
return base + bonus - penalty, after_axes, flags
def top3_candidates(prev: chess.Board) -> list[tuple[chess.Move, float, dict[str, float], list[str]]]:
me = prev.turn
scored = []
for mv in prev.legal_moves:
s, ax, fl = candidate_score(prev, mv)
scored.append((mv, s, ax, fl))
scored.sort(key=lambda x: x[1], reverse=True) # higher is better for mover (always computed from mover POV)
return scored[:3]
def format_candidate(prev: chess.Board, mv: chess.Move, before_axes: dict[str, float], s: float, after_axes: dict[str, float], flags: list[str]) -> str:
# Pick best + worst axis change for a compact explanation
top3, _all = top_deltas(before_axes, after_axes, k=3)
pos = [t for t in top3 if t[2] > 0]
neg = [t for t in top3 if t[2] < 0]
pro = axis_pretty(pos[0][0]) if pos else "Structure"
con = axis_pretty(neg[0][0]) if neg else "Low risk"
risk = flags[0] if flags else "Low immediate risk."
return (
f"- **{prev.san(mv)}** ({uci_to_label(prev, mv)})\n"
f" - Why: **{pro}** improves (score≈{s:+.2f}).\n"
f" - Risk: {risk if flags else con + '.'}"
)
# -----------------------------
# Main move handler
# -----------------------------
def build_hds_trace(prev_fen: str, new_fen: str) -> str:
prev = chess.Board(prev_fen)
mv = infer_move(prev_fen, new_fen)
# If we can't infer (board edited), just show eval snapshot
me = prev.turn
before_axes = eval_axes(prev, me)
if mv is None:
return (
"⚠️ Could not infer the move (board might be edited, not moved).\n\n"
"HDS Trace:\n\n"
+ axis_table_md(before_axes, before_axes)
)
after = chess.Board(new_fen)
after_axes = eval_axes(after, me)
top3, _all = top_deltas(before_axes, after_axes, k=3)
flags = risk_flags(prev, mv, after)
verdict1 = short_verdict(before_axes, after_axes, me)
mover = side_name(me)
move_line = f"Move: **{mover} plays {prev.san(mv)}** ({uci_to_label(prev, mv)})."
# HDS-ish trace block
trace = []
trace.append("### HDS Trace (deterministic)")
trace.append(f"- {verdict1}")
trace.append(f"- {move_line}")
trace.append(f"- {causal_sentence(top3, flags)}")
# State vector table (before/after)
trace.append("\n**State vector (axes)**")
trace.append(axis_table_md(before_axes, after_axes))
# Candidate moves (computed from the position *before* your move)
trace.append("\n### Top-3 candidate moves (with pros / risks)")
cands = top3_candidates(prev)
for (cmv, cs, cax, cfl) in cands:
trace.append(format_candidate(prev, cmv, before_axes, cs, cax, cfl))
# Keep it readable; still “HDS-ish” without garbage jargon
return "\n".join(trace)
def on_move(fen_after: str, prev_fen: str):
if not prev_fen:
prev_fen = START_FEN
md = build_hds_trace(prev_fen, fen_after)
return fen_after, fen_after, md
def on_reset():
md = (
"### HDS Trace (deterministic)\n"
"- State: Roughly equal.\n"
"- Move: (none yet)\n"
"- Goal: build a safe structure, then increase pressure.\n"
)
return START_FEN, START_FEN, md
CSS = """
footer { display: none !important; }
.gradio-container { max-width: 980px !important; }
"""
with gr.Blocks(css=CSS, title="HDS Chess Trainer") as demo:
prev_state = gr.State(START_FEN)
board = Chessboard(
value=START_FEN,
game_mode=True,
orientation="white", # fixed as requested
show_label=False,
label=None,
)
log = gr.Markdown(
value=(
"### HDS Trace (deterministic)\n"
"- State: Roughly equal.\n"
"- Move: (none yet)\n"
"- Goal: build a safe structure, then increase pressure.\n"
)
)
reset_btn = gr.Button("Reset")
board.move(on_move, inputs=[board, prev_state], outputs=[board, prev_state, log])
reset_btn.click(on_reset, inputs=None, outputs=[board, prev_state, log])
if __name__ == "__main__":
demo.launch()
|