Oct 26 2 days ago 07:02 weekend 2026-10-24/25: Ornith 1.5 122B report + board rows, sampled-AIME pilot report, csv/card-fit/depth rows Oct 25 3 days ago 07:37 sub10 addendum + toolcall-v1 (2026-10-25): dataset/README.md 07:37 sub10 addendum + toolcall-v1 (2026-10-25): dataset/matharena_contest_32k.csv 07:37 sub10 addendum + toolcall-v1 (2026-10-25): dataset/board_ci.csv 07:37 sub10 addendum + toolcall-v1 (2026-10-25): dataset/toolcall_v1.csv Oct 20 8 days ago 06:35 bonsai2 phase 2: AIME/HMMT 2026 + GPQA think-on rows Oct 19 9 days ago 15:50 Qwen3.8-27B-Coder390 vs base Q6_K: second-tier row, report, chart Oct 15 13 days ago 06:16 sub10: add Qwen3.8-27B GSQ-RCO IQ2_XS row (asked for by a reader) Oct 13 15 days ago 05:27 sub-10 GB Qwen3.8-27B: Underdog-Saluki-27B study (report, chart, second_tier.csv rows, Field Reports row) Oct 11 17 days ago 13:04 hermes-harness transfer: FineEnvs LFM2.5-2.6B checkpoints inside Hermes Agent (report, chart, csv) 12:16 second tier 16k: Ternary-Bonsai-2-27B PTQ1_0 + PQ2_0 rows and report addendum Oct 7 21 days ago 11:42 this year's contests at 32k (AIME 2026 + HMMT Feb 2026, nine think-on rows), GPQA Diamond 32k pass, plus the MATH-500 32k pass and four reports the card already linked Sep 20 09:24 second tier think-on 16k board: dataset/second_tier.csv 09:24 second tier think-on 16k board: dataset/README.md Sep 9 13:10 Upload dataset/README.md with huggingface_hub 12:49 Upload dataset/speed_at_depth.csv with huggingface_hub 12:49 Upload dataset/README.md with huggingface_hub 10:09 Upload dataset/README.md with huggingface_hub 08:41 Upload dataset/README.md with huggingface_hub 08:10 Upload dataset/README.md with huggingface_hub 06:47 Upload dataset/card_fit.csv with huggingface_hub 06:47 Upload dataset/card_fit.md with huggingface_hub 06:47 Upload dataset/README.md with huggingface_hub 06:23 Upload dataset/speed_at_depth.csv with huggingface_hub 06:23 Upload dataset/README.md with huggingface_hub 05:59 Upload dataset/benchmarks.csv with huggingface_hub 05:59 Upload dataset/README.md with huggingface_hub 
