Alexander Whitestone 78352b0a91 feat: add long-session quality test (Issue #12) ...

50-turn multi-phase conversation test that detects quality degradation
under sustained context pressure. Supports Ollama and llama-server
backends with KV cache type configuration.

Phases: code_gen -> debug -> refactor -> test -> iterate
Metrics: quality score, coherence drift, hallucinated references,
repetition ratio, prompt relevance.

Includes --compare mode for side-by-side KV type comparison.
Acceptance: run on both TurboQuant and FP16, compare results.

2026-04-13 15:16:35 -04:00

..

perplexity_results.json

feat: wikitext-2 corpus + perplexity benchmark script (closes #21)

2026-04-12 00:39:14 -04:00

prompts.json

feat: add standardized benchmarking prompts

2026-03-30 21:14:48 +00:00

run_benchmarks.py

feat: multi-backend benchmark suite with TTFT + memory tracking (#37)

2026-04-13 14:05:17 +00:00

run_long_session.py

feat: add long-session quality test (Issue #12)

2026-04-13 15:16:35 -04:00

run_perplexity.py

feat: wikitext-2 corpus + perplexity benchmark script (closes #21)

2026-04-12 00:39:14 -04:00

test_prompts.json

Add benchmark test prompts for quality comparison (Issue #22)

2026-03-31 17:31:05 +00:00