24 days ago
-
docs: Update Model Card for Veloxy-3B Zero Intelligence Loss release
-
feat: Release Veloxy-3B (Zero Intelligence Loss, Merged 18T Token Base, Aligned on A100)
-
docs & fix: Update architecture, float32 RMSNorm, on-the-fly RoPE, and comprehensive Model Card
-
docs: Publish Live Studio URL https://merchandise-computed-ivory-barrier.trycloudflare.com
-
docs: Add Veloxy-1B Pretraining Report
-
feat: Release Veloxy-1B Modern Causal Transformer Pretrained From Scratch
-
initial commit