topcatmax/mirror-unconst-affine-5czsc2fc98-r1032-vera-odpo-midrank-hibeta-shortctx-ultraextra-ep4-midlr-me

R1032 — ShortCtx × MidRank × HiBeta Ultra MidLR (offline DPO on vera king)

Affine SN120 challenger for Reason v4 (weight_version_key=7): tempered multi-sample log-mean-exp over k=3 teacher refs (τ=0.03).

Per turn: a_i = lpC(y_i|z_A) − lpC(y_i|∅); Reason = τ·log(mean_i exp(a_i/τ)). Crown also needs median stripped |z|≥80 and B pass ≥0.30.

How this checkpoint was trained

• —Base / parent: vera6/affine-5g4yy75zuz-t6@8e3f1695e058837ed80fec3238ff439fdc2d0f0e (live king reign36)
• —Method: offline DPO on Reason-ranked duel pairs (not SFT / not online GRPO)
• —What was optimized: preference for thoughts that raise teacher-side Reason (commit to a teacher next-action mode; filler loses under LME)
• —Data: Soft Mid Mid Soft → ShortCtx filtered duel preference pairs (dpo_duel_reason.jsonl, 604 lines) under mining/experiments/r1032-vera-offline-dpo-hialpha-midrank-hibeta-shortctx-ultrasuperextrasteps-ep4-midlr / pod /root/r1032/
• —Key hyperparameters:
• —LoRA r=32 (MidRank), α=128 (HiAlpha)
• —β=0.3 (HiBeta)
• —lr=1e-6 (MidLR)
• —max_len=6144 (ShortCtx)
• —max_steps=28800 (UltraSuperExtra)
• —epochs=4
• —Hardware: Lium mine-r924-vera-midctx-hibeta-1 (cosmic-orbit-55) 8×H200 GPUs 1,3 train+merge; TKC warm; chall :8003 GPUs 1,3 for v4 n80 → /tmp/r1032_merged (~16 safetensor shards)
• —Local n80 vs live king reign36 (vera6/affine-5g4yy75zuz-t6@8e3f1695e058837ed80fec3238ff439fdc2d0f0e) under wvk=7:
• —margin +0.005461, SE 0.002603, z=2.098, n=80
• —bar max(2·SE, δ=0.002) = 0.005206 (~1.049×)
• —thought median 175.5 (≥80 ✓), B pass 0.475 (≥0.30 ✓)
• —k=3, τ=0.03 (fail-closed if stamp ≠ v4)
• —decision: WIN / Stage-5 licensed (r1032_sim_result_reign36_wvk7.json, p4177)
• —Lineage: R1015 ShortCtx MidRank Hiβ Mega MidLR REFUTE m=−0.002353 ~−0.94× → ShortCtx MidRank Hiβ Ultra MidLR isolate (Mega→Ultra); ≠ Mega MidLR R1015 / ≠ Mega UltraLoLR R1001 / ≠ Ultra HiLR R984 / ≠ SoftCtx MidRank Hiβ Mega MidLR R1027 / ≠ MidCtx MidRank Hiβ Ultra MidLR R1017 / ≠ Online / ≠ GRPO
• —Experiment path: mining/experiments/r1032-vera-offline-dpo-hialpha-midrank-hibeta-shortctx-ultrasuperextrasteps-ep4-midlr

Intended use

SN120 Affine miner submission / evalsrv Reason v4 duel. Not a general chat model.

License

Follows base model + Affine mining artifacts policy.