evaluation 12
- Contamination Checks Before You Trust a Fine-Tune Score
- Small AI by design: ten Kenyan startups and the offline-first constraint
- Fuse Ranks, Not Scores: Combining BM25 and Dense Retrieval
- Is Your +3-Point Eval Gain Real? Gating a Comparison with Paired Statistics
- Twelve Hours and 1,239 Certificates: What Kenya's Civil-Service AI Course Delivered
- The Cache Was Confident and Wrong: Choosing a Semantic Cache Threshold You Can Defend
- Agreement Before Accuracy: Auditing the Labels You Are About to Train On
- Your Corpus Is Duplicating Itself — MinHash + LSH Dedup You Can Measure Tonight
- The $1 Billion AI Bet: What the Goalkeepers Report Funds, and How to Read Its Four Numbers
- The Mask Is the Contract: What Grammar-Constrained Decoding Actually Guarantees
- Evaluating LLMs: Benchmarks That Matter for African Use Cases
- Automated AI Red Teaming at Scale