bench-spec-ab: think-free echo/structured shapes, and prove the echo is an echo

The first run asked for a verbatim repeat with thinking ON, so the model copied
the passage inside reasoning_content, content came back empty for all three
configurations, and the cross-configuration identity check compared '' to '' --
a pass that proves nothing. echo and structured now disable thinking via
chat_template_kwargs, the echo row carries a copy_ratio (share of answer lines
that really are passage lines), and an empty answer in any shape is a loud
warning rather than a silent pass.

'structured' is new and answers the case the fleet actually runs on: repetitive
OUTPUT with no prompt overlap (record lists, tables, config rewrites), which is
where an n-gram matcher should pay and an echo-shaped test cannot tell.

scripts/bench-spec-ab.py

@@ -17,25 +17,28 @@ Config syntax: `name=SPEC_TYPE[@n_max]`, semicolon-separated. The systemd unit i
 restored at the end, so the endpoint comes back on whatever the env file says
 whatever happens here.
 
-TWO SHAPES, because a single one answers the wrong question:
+THREE SHAPES, because a single one answers the wrong question. A proposer can
+only draft text it has seen, so the shape decides the verdict:
 
-  prose  a 1k-token prompt asking for a sentence — nothing to copy, so only a
+  prose       a 1k prompt asking for one sentence. Nothing to copy and nothing
-         model-backed proposer can help. This is the shape that compares
+              repetitive, so only a model-backed proposer can help: this is the
-         drafters.
+              shape that compares drafters.
-  echo   a ~1.5k-token passage of unique lines with "repeat it back verbatim".
+  echo        a ~1.5k passage of unique lines, "repeat it back verbatim". The
-         The output is almost entirely a copy of the prompt, which is the only
+              output IS the prompt, which is ngram-mod's home turf.
-         traffic where an n-gram matcher has anything to match. Measuring
+  structured  "25 JSON records, one per line" — repetitive OUTPUT with no prompt
-         ngram-mod on prose alone would call it useless; measuring DFlash on
+              overlap, which is what an ngram matcher is really for in an
-         echo alone would call it wasteful.
+              agentic fleet (config edits, tables, record lists).
 
-Results on Dankserver (frank-medium payload, 2026-09-16, temperature 0):
+echo and structured disable thinking (`chat_template_kwargs`), otherwise the
+model thinks the answer into `reasoning_content`, `content` comes back empty and
+the cross-configuration identity check passes vacuously — which is exactly how
+this harness first "verified" an echo it never looked at.
 
-    prose 1k/512      none 17.7 | mtp n=3 32.4 | mtp n=7 26.7 | dflash=7 43.5 tok/s
+Results on Dankserver (frank-medium payload, 2026-09-16, temperature 0). prose
-    card 1 VRAM       none 26.07 | mtp n=3 29.35 | mtp n=7 29.65 | dflash 29.42 GB
+1k/512 decode: none 17.7 | mtp n=3 32.4 | mtp n=7 26.7 | dflash=7 43.5 tok/s.
-
+`draft-mtp,draft-dflash` fails at load (a chain has ONE draft context and MTP
-...and `draft-mtp,draft-dflash` fails at load: a chain has ONE draft context and
+builds against whatever model that context holds); ngram-mod has no such
-MTP builds against whatever model that context holds. ngram-mod has no such
+constraint — it is draftless and joins the chain unconditionally.
-constraint — it is draftless and is added to the chain unconditionally.
 """
 import argparse
 import json
@@ -51,7 +54,8 @@ FILLER = ("The quick brown fox jumps over the lazy dog near the river bank. "
 TOKENS_PER_REP = 28
 UNIT = "frank-medium.service"
 MODEL = "frank-medium"
-SPEC_GREP = "adding speculative implementation|- n_max=|- block_size=|creating MTP|ngram_mod with"
+SPEC_GREP = ("adding speculative implementation|- n_max=|- block_size=|creating MTP|"
+             "draft acceptance")
 
 DEFAULT_CONFIGS = [
     ("none", {"SPEC_TYPE": "none"}),
@@ -73,10 +77,14 @@ def echo_passage(lines=90):
         for i in range(lines))
 
 
-ECHO_INSTRUCTION = ("\n\nRepeat the passage above verbatim, every line, exactly as written, "
+ECHO = (echo_passage() + "\n\nRepeat the passage above verbatim, every line, exactly as "
-                    "with no commentary before or after it.")
+        "written, with no commentary before or after it.")
-PROSE_INSTRUCTION = "\n\nReply with a single short sentence."
+STRUCTURED = ("Output exactly 25 JSON records, one per line, no prose and no code fence. "
-IDENTITY_PROMPT = "List the first five prime numbers, comma separated, then stop."
+              'Each line must look like {"id": N, "state": "active", "region": "us-west-2"} '
+              "with N running from 1 to 25.")
+PROSE = FILLER * (1000 // TOKENS_PER_REP) + "\n\nReply with a single short sentence."
+IDENTITY = "List the first five prime numbers, comma separated, then stop."
+NO_THINK = {"chat_template_kwargs": {"enable_thinking": False}}
 
 
 def log(msg, logfile):
@@ -159,11 +167,15 @@ def stop_unit():
         time.sleep(1)
 
 
-def run_shape(base, prompt, max_tokens, label):
+def run_shape(base, prompt, max_tokens, label, extra=None):
-    t = post(base, {"model": MODEL,
+    body = {"model": MODEL, "messages": [{"role": "user", "content": prompt}],
-                    "messages": [{"role": "user", "content": prompt}],
+            "max_tokens": max_tokens, "temperature": 0.0}
-                    "max_tokens": max_tokens, "temperature": 0.0})
+    if extra:
-    tm = t["timings"]
+        body.update(extra)
+    d = post(base, body)
+    tm = d["timings"]
+    msg = d["choices"][0]["message"]
+    content = (msg.get("content") or "").strip()
     row = {"shape": label,
            "prompt_n": tm["prompt_n"],
            "prefill_tps": round(1000 * tm["prompt_n"] / tm["prompt_ms"], 1),
@@ -171,18 +183,26 @@ def run_shape(base, prompt, max_tokens, label):
            "decode_tps": round(tm["predicted_per_second"], 2),
            "draft_n": tm.get("draft_n", 0),
            "accepted": tm.get("draft_n_accepted", 0),
-           "answer": (t["choices"][0]["message"].get("content") or "").strip()}
+           "answer": content,
+           "reasoning_len": len(msg.get("reasoning_content") or "")}
+    if label.startswith("echo"):
+        # How much of the answer really is the passage: a config that "wins" by
+        # answering something else is not a win.
+        want = {l.strip() for l in echo_passage().splitlines()}
+        got = [l.strip() for l in content.splitlines() if l.strip()]
+        row["copy_ratio"] = round(sum(1 for l in got if l in want) / len(got), 3) if got else 0.0
     return row
 
 
 def measure(base):
-    run_shape(base, FILLER * (4000 // TOKENS_PER_REP) + PROSE_INSTRUCTION, 32, "warmup")
+    run_shape(base, FILLER * (4000 // TOKENS_PER_REP) +
-    rows = [
+              "\n\nReply with a single short sentence.", 32, "warmup")
-        run_shape(base, FILLER * (1000 // TOKENS_PER_REP) + PROSE_INSTRUCTION, 512, "prose 1k/512"),
+    return [
-        run_shape(base, echo_passage() + ECHO_INSTRUCTION, 400, "echo 1.5k/400"),
+        run_shape(base, PROSE, 512, "prose 1k/512"),
-        run_shape(base, IDENTITY_PROMPT, 192, "identity"),
+        run_shape(base, ECHO, 400, "echo 1.5k/400", NO_THINK),
+        run_shape(base, STRUCTURED, 400, "structured/400", NO_THINK),
+        run_shape(base, IDENTITY, 192, "identity"),
     ]
-    return rows
 
 
 def main():
@@ -219,22 +239,26 @@ def main():
             continue
         rows = measure(base)
         results[name] = {"vram_gb": vram(), "rows": rows,
-                         "spec_log": sh("grep -E '" + SPEC_GREP + "' " + logf + " | head -6")}
+                         "spec_log": sh("grep -E '" + SPEC_GREP + "' " + logf + " | tail -4")}
         for r in rows:
             log(name + " | " + r["shape"] + " prefill " + str(r["prefill_tps"]) +
                 " tok/s | decode " + str(r["decode_tps"]) + " tok/s | draft " +
                 str(r["draft_n"]) + "/" + str(r["accepted"]) +
-                (" (" + str(round(100.0 * r["accepted"] / r["draft_n"])) + " %)" if r["draft_n"] else ""),
+                (" (" + str(round(100.0 * r["accepted"] / r["draft_n"])) + " %)" if r["draft_n"] else "") +
-                logfile)
+                (" | copy " + str(r["copy_ratio"]) if "copy_ratio" in r else "") +
+                (" | EMPTY ANSWER" if not r["answer"] else ""), logfile)
         log(name + " | vram " + str(results[name]["vram_gb"]), logfile)
         stop_unit()
 
     with open(args.json, "w") as fh:
         json.dump(results, fh, indent=2)
-    for shape in ("prose 1k/512", "echo 1.5k/400", "identity"):
+    for shape in ("prose 1k/512", "echo 1.5k/400", "structured/400", "identity"):
-        texts = {v["rows"][i]["answer"] for v in results.values() if "rows" in v
+        texts = [r["answer"] for v in results.values() if "rows" in v
-                 for i, r in enumerate(v["rows"]) if r["shape"] == shape}
+                 for r in v["rows"] if r["shape"] == shape]
-        if len(texts) > 1:
+        if any(not t for t in texts):
+            log("WARNING: '" + shape + "' produced an empty answer in some configuration -- "
+                "its identity check proves nothing", logfile)
+        elif len(set(texts)) > 1:
             log("WARNING: configurations disagree on greedy text for '" + shape +
                 "' -- speculation is changing answers", logfile)
     log("restoring " + UNIT, logfile)