minesweeper
mixed_sdft_solution_minesweeper_qwen3_4b_thinking_1_epoch_8192_32_batch_2e-5_lr_nemotron8bmixed_sdft_solution_sequential_minesweeper_kukurasu_qwen3_4b_thinking
Model Card for mixed_sdft_solution_sequential_minesweeper_kukurasu_qwen3_4b_thinking_1_epoch_8192_32_batch_2e-5_lr_qwen3_1_7b
This model is a fine-tuned version of None.
It has been trained using TRL.
Quick start
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="None", device="cuda")
output =… See the full description on the dataset page: https://huggingface.co/datasets/SeanWang0027/mixed_sdft_solution_sequential_minesweeper_kukurasu_qwen3_4b_thinking.minesweeper-solver-benchmark
Minesweeper Solver Benchmark — Win-Rate by Difficulty
Full-game win-rate for the LK Forge Minesweeper Solver
across the three standard difficulties. The shipped engine turns every revealed number into
a constraint and applies counting → the subset rule → exact enumeration of all valid mine
arrangements; when logic is exhausted it opens the hidden cell with the lowest computed mine
probability. 1,000 seeded games per difficulty. Rules: classic board sizes, the first click is… See the full description on the dataset page: https://huggingface.co/datasets/LKForge/minesweeper-solver-benchmark.mixed_sdft_solution_minesweeper_qwen3_4b_thinking_1_epoch_8192_32_batch_2e-5_lr_qwen3_1_7bMinesweeper_offlineminesweeper_taskset