Full BF16 checkpoint — 453 tensors, 6.155B parameters , no NaN / no Inf. Clean build.
▸ layers.* (transformer blocks) — 390 tensors · 5.43 B params · 88.2% ▸ noise_refiner — 26 tensors · 361.8 M params · 5.9% ▸ context_refiner — 22 tensors · 353.9 M params · 5.8% ▸ cap_embedder — 3 tensors · 9.8 M params · 0.2% ▸ final_layer — 4 tensors · 1.2 M params · t_embedder — 4 tensors · 0.5 M params · x_embedder — 2 tensors · 0.25 M params · Total: 453 tensors · 6.155 B parameters
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📊 Weight Distribution
▸ Global range: min ≈ −14.00 , max ≈ +13.94 — in line with typical DiT checkpoints
▸ Most active modules (highest std): the deep layers layers.26 through layers.29 — this is where the style adjustment is concentrated. Their ffn_norm2 and attention_norm2 tensors show std up to 3.2 vs. a model average of ~ 0.32
▸ Most conservative modules: t_embedder (std ~0.005–0.02) — timestep embedding is nearly untouched, as expected from a weight-adjustment rather than a retrain
▸ Feed-forward w2 weights carry the largest absolute values (up to ±14), consistent with how Z-Image's MLP projections store learned priors
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
✅ File Integrity
▸ NaN tensors: 0 ▸ Inf tensors: 0 ▸ Dtype consistency: 100% BF16 ▸ Architecture match vs. z_image_turbo_bf16 : structurally identical (906/906 keys)
