Logic pass: CPU attention mirrors get the same int32 bound as the GPU kernels (peers refuse identically); gpuMlpChain returns pooled buffers when the row-max guard throws 4060984 verified Quazim0t0 commited on 10 days ago
Reductions refuse what they used to accept: GPU row-max non-finite guard, averageGrads length/NaN refusal, int32 accumulator bound, capacity-weighted cluster loss, finite reduced-gradient check, empty CUDA_VISIBLE_DEVICES 6e87247 verified Quazim0t0 commited on 10 days ago
web: restore LF line endings on verified_core.js / transformer.js 28383f7 verified Quazim0t0 commited on 13 days ago
Refuse NaN/Inf at every int8 quantize; requant rounds half-up (retrained); bounded units; README for September 99509b7 verified Quazim0t0 commited on 13 days ago
Upload web/public/verified_core.js with huggingface_hub 7fcb6bd verified Quazim0t0 commited on Jul 18
Upload web/public/verified_core.js with huggingface_hub d5a32d6 verified Quazim0t0 commited on Jul 18
Upload web/public/verified_core.js with huggingface_hub 75f9477 verified Quazim0t0 commited on Jul 17
web: gate-the-gate boundary discrimination in the B2B init gate b23b595 verified Quazim0t0 commited on Jul 16
web: logic-pass fixes (portable test_b2b, symmetric repair patience) b004b80 verified Quazim0t0 commited on Jul 16
web: B2B MLP chain (CUTLASS ex. 13 two-GEMM fusion + ex. 23 epilogue reduction), test_b2b.js, ten-suite npm test 4a8b60f verified Quazim0t0 commited on Jul 16
web: QKV dual-GEMM fusion (bit-identical, CUTLASS ex. 45) 30eb3fa verified Quazim0t0 commited on Jul 16
web: sync-guard gradient repair, FineWeb-Edu hardcoded (parquet /data streaming), dispatch-optimized backward (bit-identical), IEEE-754 oracle + corpus mutation tests, TEST_RESULTS.md 9ae1216 verified Quazim0t0 commited on Jul 16
Web demo: lr default 0.01, batched per-head backward GEMMs 3dbd21f verified Quazim0t0 commited on Jul 15
Web demo: exact kernel gates + continuous cross-device kernel verification 93d9b2e verified Quazim0t0 commited on Jul 15
Web demo: weight tying + WebRTC resilience (relay fallback, keepalive) 00fdb1d verified Quazim0t0 commited on Jul 14
Web demo: block-scaled INT8 + batched/fused GEMM kernels 32f59c2 verified Quazim0t0 commited on Jul 14
Web demo: leader failover + dataset picker + reconnect policy 56e2384 verified Quazim0t0 commited on Jul 13
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 4cf5cdc verified Quazim0t0 commited on Jul 13
Web demo: room-first lobby + group training settings sliders 4bb3f7c verified Quazim0t0 commited on Jul 12
Release: SpikeWhale slider panel (HF dataset picker, stop/back), DaisyChain-Web (P2P WebRTC training, DaisyAdam, checkpoints, room host approval, verified-units-only) 4fd620e verified Quazim0t0 commited on Jul 12
