Web demo: lr default 0.01, batched per-head backward GEMMs

Web demo: metamorphic property tests (no-oracle correctness checks)

Web demo: exact kernel gates + continuous cross-device kernel verification

Web demo: split-K backward + gather-fused attention

Web demo: weight tying + WebRTC resilience (relay fallback, keepalive)

Web demo: block-scaled INT8 + batched/fused GEMM kernels

Web demo: leader failover + dataset picker + reconnect policy

Web demo: mid-run joins + auto-reconnect

Web demo: multi-device sync + connection fixes

Web trainer upgrade: sync guard (leader roster + weight-hash divergence check), 3xINT8 fast-accurate GEMM, DP4A hardware path (verified vs units), Spikewhale tokenizer (16.5k vocab), FineWeb-Edu streaming, gradient/checkpoint fragmentation, inference kit, generation tester, contribution logs

Web demo: mini transformer LM + scaling sliders

Web demo: room-first lobby + group training settings sliders

Release: SpikeWhale slider panel (HF dataset picker, stop/back), DaisyChain-Web (P2P WebRTC training, DaisyAdam, checkpoints, room host approval, verified-units-only)