frank-default-small: ubatch=1536, ctx-checkpoints=2 (24GB VRAM safety)

scripts/frank-default-small.env

@@ -41,11 +41,11 @@ CACHE_TYPE_V=f16
 
 VERBOSITY=4
 # Store SWA/context checkpoints in RAM to hold peak GPU memory down.
-CTX_CHECKPOINTS=4
+CTX_CHECKPOINTS=2
 
-# Big prefill batches (ubatch 1024, batch 4096).
+# Big prefill batches (ubatch 1536, batch 4096).
 BATCH_SIZE=4096
-UBATCH_SIZE=1024
+UBATCH_SIZE=1536
 
 # Eager RAM load (MoE experts on CPU; ~31 GB RAM).
 NO_MMAP=1