diag: add late-layer diagnostics, fix ffn ctx variable

single_shot_inference.py

@@ -513,7 +513,7 @@ def forward_layer(X_l, w, li, cfg, rope_cos, rope_sin,
     # -- mHC post_block (attention) --
     X_mid = attn_mhc.post_block(X_l, F_attn, attn_ctx)  # (T, n_hc, H)
     # Diagnostic: check mHC is stabilizing the residual
-    if li < 3:
+    if li < 3 or li >= n_layers - 3:
         B_l, C_l = attn_ctx
         print(f"  L{li} attn: |X_l|={X_l.abs().max().item():.2f} |F_attn|={F_attn.abs().max().item():.2f} |B|={B_l.abs().max().item():.4f} |C|={C_l.abs().max().item():.4f} |X_mid|={X_mid.abs().max().item():.2f}", flush=True)
 
@@ -532,9 +532,9 @@ def forward_layer(X_l, w, li, cfg, rope_cos, rope_sin,
 
     # -- mHC post_block (FFN) --
     X_next = ffn_mhc.post_block(X_mid, F_ffn, ffn_ctx)  # (T, n_hc, H)
-    if li < 3:
-        B_l, C_l = ffn_ctx
-        print(f"  L{li} ffn: |X_mid|={X_mid.abs().max().item():.2f} |F_ffn|={F_ffn.abs().max().item():.2f} |B|={B_l.abs().max().item():.4f} |C|={C_l.abs().max().item():.4f} |X_next|={X_next.abs().max().item():.2f}", flush=True)
+    if li < 3 or li >= n_layers - 3:
+        B_l_ffn, C_l_ffn = ffn_ctx
+        print(f"  L{li} ffn: |X_mid|={X_mid.abs().max().item():.2f} |F_ffn|={F_ffn.abs().max().item():.2f} |B|={B_l_ffn.abs().max().item():.4f} |C|={C_l_ffn.abs().max().item():.4f} |X_next|={X_next.abs().max().item():.2f}", flush=True)
 
     return X_next