[NV] [PR2] feat(synthetic-al): add vLLM synthetic-acceptance backend / [NV] [PR2] feat(synthetic-al)：添加 vLLM 合成接受率后端

Add the vLLM recipe rewrite backend for controlled synthetic acceptance. / 添加用于受控合成接受率的 vLLM 配方重写后端。

Author: qiching

runners/synthetic_injectors/__init__.py

@@ -19,9 +19,8 @@
 """
 
 # framework value passed by the runner (e.g. "dynamo-vllm") -> backend module.
-# Populated by backend modules registering themselves (see e.g. vllm.py, added
-# in the framework-support PR). Empty here: the generic layer alone injects
-# nothing, so merging it changes no behavior.
+# Populated by backend modules registering themselves at import time (see the
+# `from . import` block below).
 _INJECTORS = {}
 
 
@@ -32,3 +31,8 @@ def register(framework, backend):
 def get_injector(framework):
     """Return the backend for ``framework`` (as passed by the runner), or None."""
     return _INJECTORS.get(framework)
+
+
+# Import backends after register/get_injector are defined so each module can
+# call register() at import time. Add new frameworks (sglang, trtllm, ...) here.
+from . import vllm  # noqa: E402,F401

runners/synthetic_injectors/vllm.py

@@ -0,0 +1,69 @@
+"""vLLM synthetic-acceptance backend (FRAMEWORK=dynamo-vllm).
+
+Rewrites every ``speculative-config: '<json>'`` entry in an srt-slurm recipe to
+use synthetic rejection sampling: it adds ``rejection_sample_method=synthetic``
+and ``synthetic_acceptance_length=<al>`` to the JSON so the engine emits a
+controlled mean acceptance length instead of running the real draft model.
+
+Registered under the "dynamo-vllm" framework key at import time, so importing
+the ``synthetic_injectors`` package is enough for the generic driver to resolve
+this backend.
+"""
+
+import json
+import re
+import sys
+
+from . import register
+
+# Matches `speculative-config: '<json>'` (single-quoted JSON, as written in the
+# recipe YAML). Capturing the JSON lets us edit it with the json module instead
+# of string-munging, so we never produce malformed quoting.
+_SPEC_CONFIG_RE = re.compile(r"speculative-config:\s*'([^']+)'")
+
+
+def spec_tokens_from_recipe(text):
+    """Best-effort: read num_speculative_tokens from the recipe itself."""
+    for m in _SPEC_CONFIG_RE.finditer(text):
+        try:
+            spec = json.loads(m.group(1))
+        except json.JSONDecodeError:
+            continue
+        n = spec.get("num_speculative_tokens")
+        if n:
+            return int(n)
+    return 2
+
+
+def rewrite(content, al, log):
+    """Rewrite every speculative-config entry to synthetic acceptance.
+
+    Returns ``(new_content, count)`` where count is the number of entries
+    modified (0 => nothing matched, recipe left unchanged by the driver).
+    """
+    before = [ln.strip() for ln in content.splitlines() if _SPEC_CONFIG_RE.search(ln)]
+    if before:
+        log("Before:")
+        for ln in before:
+            print(f"  {ln}")
+
+    def _replace(match):
+        spec = json.loads(match.group(1))
+        spec["rejection_sample_method"] = "synthetic"
+        spec["synthetic_acceptance_length"] = al
+        # Compact separators keep the same style as the hand-written recipes.
+        return "speculative-config: '" + json.dumps(spec, separators=(",", ":")) + "'"
+
+    new_content, count = _SPEC_CONFIG_RE.subn(_replace, content)
+
+    if count:
+        after = [ln.strip() for ln in new_content.splitlines() if _SPEC_CONFIG_RE.search(ln)]
+        if after:
+            log("After:")
+            for ln in after:
+                print(f"  {ln}")
+
+    return new_content, count
+
+
+register("dynamo-vllm", sys.modules[__name__])