TildAlice – Tech Blog for Developers
Python, AI, Deep Learning, Data Analysis, Computer Vision, Reinforcement Learning โ tips and tutorials
Latest Posts
-
setup.py to pyproject.toml: PEP 621 Migration Guide
Migrate from setup.py to pyproject.toml using PEP 621 standards. Fix dependency conflicts and streamline your Python packaging workflow.
-
Docker Build Cache: 7 Layer Patterns That Cut CI by 73%
Cut Docker CI builds from 14min to 47sec with 7 layer caching patterns. Real measurements on dependency ordering, BuildKit mounts, and multi-stage builds.
-
GPU vs CPU Inference: 5 Scenarios, Real Costs & Latency
GPU vs CPU inference across 5 traffic scenarios: real costs, latency benchmarks, and when each makes sense. BERT/ResNet/Whisper tested on AWS.
-
Greedy vs DP: When Sorting Fails the Interview
Learn when greedy algorithms fail and dynamic programming saves the day through a classic interval scheduling problem that trips up interviews
-
Mamba vs RWKV: 32K Context Benchmark on A100
Mamba vs RWKV: real accuracy and memory numbers at 32K tokens on A100. One architecture chokes past 16K โ the other scales but misses facts.
-
Pandas Memory Optimization: Cut RAM Usage by 90% on Large CSVs
Optimize Pandas memory usage and cut RAM by 90% when loading large CSVs. Learn dtype tricks, chunking, and category conversion techniques.
-
EfficientNetV2 vs ResNet: 11x Faster Training Explained
EfficientNetV2 trains 11x faster than ResNet through progressive learning, Fused-MBConv blocks, and adaptive regularization strategies.