context-compressor
Public
bench / results / 2026-08-24-gemma-3-12b-it-heretic.md
bench / results / 2026-08-24-gemma-3-12b-it-heretic.md
Model: gemma-3-12b-it-heretic
Settings: {"limitTokens":500,"floorTokens":400,"keepRecentTokens":300,"chunkTokens":800,"summarizerMaxTokens":400,"consolidationBudgetTokens":600,"consolidationKeepNewest":2,"mergeMaxTokens":400,"maxExcerptChars":32000,"quizMaxTokens":120}
| Fixture | Recall (compacted) | Recall (consolidated) | Reduction | Chunk latency mean/p50/max (ms) | Structure | Cache |
|---|---|---|---|---|---|---|
| coding-agent | 5/5 | — | 43% | 3224.3/2885/4441 | ✅ | ✅ |
| research | 3/3 | — | 40% | 3403/3403/3403 | ✅ | ✅ |
| contradictory-updates | 3/3 | — | 31% | 1497/1497/1497 | ✅ | ✅ |
| tool-spam | 0/3 | 0/3 | 63% | 1829.5/1698/3363 | ✅ | ✅ |
Recall = seeded facts answered correctly from the compacted view alone. Results are comparable only within one model; see bench/README.md.
Model: gemma-3-12b-it-heretic
Settings: {"limitTokens":500,"floorTokens":400,"keepRecentTokens":300,"chunkTokens":800,"summarizerMaxTokens":400,"consolidationBudgetTokens":600,"consolidationKeepNewest":2,"mergeMaxTokens":400,"maxExcerptChars":32000,"quizMaxTokens":120}
| Fixture | Recall (compacted) | Recall (consolidated) | Reduction | Chunk latency mean/p50/max (ms) | Structure | Cache |
|---|---|---|---|---|---|---|
| coding-agent | 5/5 | — | 43% | 3224.3/2885/4441 | ✅ | ✅ |
| research | 3/3 | — | 40% | 3403/3403/3403 | ✅ | ✅ |
| contradictory-updates | 3/3 | — | 31% | 1497/1497/1497 | ✅ | ✅ |
| tool-spam | 0/3 | 0/3 | 63% | 1829.5/1698/3363 | ✅ | ✅ |
Recall = seeded facts answered correctly from the compacted view alone. Results are comparable only within one model; see bench/README.md.