diff --git a/single_shot_inference.py b/single_shot_inference.py index 5fbfd9dc..535673d3 100644 --- a/single_shot_inference.py +++ b/single_shot_inference.py @@ -61,7 +61,7 @@ PROMPT = _args.prompt or "The capital of France is" NUM_GPUS = 8 SKIP_ROUTED_MOE = _args.skip_moe # If True, only use shared expert (debug) INVERSE_ROPE = not _args.no_inverse_rope # If False, skip inverse RoPE on attention output (diagnostic) -MHC_DIAG = False # If True, print per-layer mHC diagnostics (B_l row/col sums, C_l values) +MHC_DIAG = True # If True, print per-layer mHC diagnostics (B_l row/col sums, C_l values) # When True: applies inverse RoPE at query position → converts absolute→relative # When False: leaves relative position encoding intact for output projection # DSV4 partial RoPE only affects last 64/512 dims; first 448 are always un-RoPE'd