TildAlice – Tech Blog for Developers
Python, AI, Deep Learning, Data Analysis, Computer Vision, Reinforcement Learning โ tips and tutorials
Latest Posts
-
๋์ ๊ณํ๋ฒ(DP) ์ฝ๋ฉํ ์คํธ ์์ ์ ๋ณต: ๊ฐ๋ ๋ถํฐ ์ค์ ๋ฌธ์ ๊น์ง
Master dynamic programming for coding interviews: derive recurrence relations, optimize space complexity, solve knapsack and LIS in Python.
-
NeurIPS 2025 ์ฃผ๋ชฉํ ๋งํ ๋ ผ๋ฌธ: Scalable Deep RL – ๋๊ท๋ชจ ๊ฐํํ์ต์ ์๋ก์ด ์งํ
NeurIPS 2025 Best Paper candidate achieves 3-4x faster deep RL training with Actor-Learner architecture and Adaptive V-trace. Implementation guide.
-
RoPE vs Alibi vs xPos: Transformer ์์น ์ธ์ฝ๋ฉ ์๋ฒฝ ๋น๊ต ๊ฐ์ด๋ (๊ธด ๋ฌธ๋งฅ ์ฒ๋ฆฌ ์ต์ ํ)
RoPE vs Alibi vs xPos on 128k token contexts: which Transformer position encoding prevents attention dilution? Benchmarks and PyTorch code inside.
-
ISO 13374 ํ์ค ๊ธฐ๋ฐ PHM ์์คํ ์ค๊ณ: OSA-CBM 6๋จ๊ณ ์ํคํ ์ฒ๋ก ๊ตฌํํ๋ ์ฐ์ ํ์ค ์์ง์ ๋น ํ๋ซํผ
Build ISO 13374-compliant PHM systems with OSA-CBM's 6-layer architecture. Python code for data acquisition to decision support included.
-
Python ๋ฐ์ฝ๋ ์ดํฐ ์์ ์ ๋ณต: @property๋ถํฐ ์ปค์คํ ๋ฐ์ฝ๋ ์ดํฐ๊น์ง ์ค์ ํ์ฉ ๊ฐ์ด๋
Fix the 3 decorator mistakes that break Flask middleware. From @property to custom decorators with performance benchmarks and caching patterns.
-
Actor-Critic ๋ฐฉ์ ์์ ์ ๋ณต: A2C๋ถํฐ SAC๊น์ง ์ฅ๋จ์ ๋น๊ต์ ํ์ดํผํ๋ผ๋ฏธํฐ ํ๋ ์ค์ ๊ฐ์ด๋
A2C vs PPO vs SAC: real training curves from 5 environments. Hyperparameter tuning secrets that cut training time 40% with Stable-Baselines3.
-
ํฌ ํฌ์ธํฐ(Two Pointers) ์๊ณ ๋ฆฌ์ฆ ์๋ฒฝ ๊ฐ์ด๋ – ๋ฐฐ์ด ํ์ O(n) ์ต์ ํ์ ์ ์
Two Pointers drops array search from O(nยฒ) to O(n). Master same-direction and opposite-direction patterns with 6 LeetCode problems solved.
-
Mamba ๋ฆฌ๋ทฐ: ์ ํ ์๊ฐ๋ณต์ก๋ State Space Model
Mamba hits O(L) complexity vs Transformer's O(Lยฒ). Selective State Space Model achieves 5ร faster inference on DNA and audio with benchmarks.