H2O Lightning 4B Long Record - Overall - Selector BM25 selection [default]: leaderboard

H2O Lightning 4B Long Record - Overall - Selector BM25 selection [default]: Correct decisions on 331 items padded to 60000 tokens (%). H2O-Lightning-4B official card: trained v1.4.1 weights (same weights served by v1.4.2), distinct from older trained v1.4.0 (additional training data documented). All current native runs retain v1.4.2 server, two-way one-engine vLLM0.30.0, calibration and profile, with explicit one-way/image-temperature variants. Owner-run full-text DI0.3 profile/default, JevBench counts, held-out and new-domain skill, text/vision indices, ECE and long-record-selector experiments retained. Skill is chance-rescaled accuracy (chance0/perfect100), with no finite theoretical floor. Copied BenchmarkHeaven ImageJevBench/resource composites and training-set measurements withheld. Native source condition: Selector BM25 selection [default].

Metric: Correct decisions on 331 items padded to 60000 tokens (%). Source: huggingface.co. 1 models tracked.

Top models

Interactive version: theaggregate.ai/benchmark?slug=h2o-lightning-4b-long-record-overall-selector-bm25-selection-default · How It Works · Data refreshed daily, snapshot 2026-11-12.