TildAlice – Tech Blog for Developers
Python, AI, Deep Learning, Data Analysis, Computer Vision, Reinforcement Learning โ tips and tutorials
Latest Posts
-
TFLite vs ONNX Mobile: 5 ARM Devices, 12ms Gap
TFLite vs ONNX benchmark on 5 ARM devices reveals 12ms latency gapโone framework dominates power efficiency while the other surprises on speed
-
Ring Attention: Train 1M Tokens on 8GB GPUs in 2026
Train transformers with 1M+ tokens on consumer GPUs using Ring Attention's distributed sequence processing. Learn the math behind blockwise compute.
-
Python GIL vs Free-Threading: 3.13t CPU Benchmark
Python 3.13t free-threading delivers 5.6x CPU speedup vs GILโbut single-threaded code is 17% slower. Real benchmark results and when to switch.
-
Pandas GroupBy 10x Faster: Map-Reduce for 100M+ Rows
Master Pandas GroupBy optimization with map-reduce patterns. Learn chunked processing and parallel techniques that handle 100M+ rows efficiently.
-
OpenAI to Claude API Migration: 7 Breaking Changes
Migrate OpenAI to Claude API: discover 7 critical breaking changes in streaming, function calling, and token handling you must fix now.
-
INT8 vs FP16 Inference: TCO Cut 54% for 7B Models on AWS
INT8 quantization slashes AWS inference costs 54% vs FP16 for 7B LLMs. Real g5.xlarge benchmarks reveal the accuracy-speed-cost tradeoffs.
-
BERT vs RoBERTa vs DistilBERT: GLUE Scores Decoded
RoBERTa beat BERT by 4+ GLUE points with zero architecture changes. Here's what actually changedโand when DistilBERT's 60% speedup is worth the tradeoff.
-
Mixture-of-Depths: Dynamic Token Skip Cuts 40% FLOPs
MoD routing lets tokens skip layers dynamically, cutting 40% FLOPs in GPT-scale models without quality loss. Here's how it works and when it breaks.