♟️ Kopernikus Chessmaster 32B (V10.0 — Self-Evolving Liquid bfloat16 + 60-Category Dual Memory)

  • Base Model: deepseek-ai/DeepSeek-R1-Distill-Qwen-32B (Full bfloat16 Liquid Training + Dual Memory Architecture)
  • Sealed Champion Engine: ik0158 — Exam Record: 43.29/100 (Tur #17)
  • Stockfish Arena Ladder: Current Max Level Reached: Lv.3/20 | Arena Score: 65W - 98D - 44L

📊 11-Pillar Live Report Card (Karne)

Category / Pillar Score (%)
Opening (acilis) %54.5
Middlegame (orta_oyun) %73.0
Puzzles & Mates (bulmaca_mat) %90.0
Syzygy Endgame TB (oyun_sonu_tb) %100.0
Reverse-Threat Safety (zorlamasiz_guvenlik) %80.0
🗡️ Piece Sacrifices (ozgun_feda 01-10) %50.0
👑 5-10 Move Deep Mates & 20 Patterns (derin_mat_10 11-26) %66.7
🛡️ High-Level Defense & Counter-Check (ust_savunma 27-36) %83.3
⚡ Tempo Gain / Zwischenzug / Desperado (tempo_kazanci 37-44) %66.7
📐 Tactical Geometry (taktik_geometri 45-52) %83.3
♟️ Domination & Underpromotion (konumsal_son 53-60) %66.7

🚀 Quick Usage (YARIN_YUKLE.py)

from YARIN_YUKLE import kopernikus_32b_ve_ikili_yukle
model, tok, cephane, oz_motor = kopernikus_32b_ve_ikili_yukle()
best_uci = oz_motor.oz_hamle_motoru("r1bqkbnr/pppp1ppp/2n5/4p3/4P3/5N2/PPPP1PPP/RNBQKB1R w KQkq - 2 3")
print("Best move:", best_uci)
Downloads last month
-
Safetensors
Model size
33B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Toramannnn/Kopernikus_chessmaster

Finetuned
(91)
this model