OpenAI API Instrumentation (e.g. seed, logprobs, echo, an array prompt)

c/tests/fixtures/captured_lmeval_request.json

@@ -0,0 +1,132 @@
+{
+  "method": "POST",
+  "path": "/v1/completions",
+  "headers": {
+    "Host": "127.0.0.1:8812",
+    "User-Agent": "python-requests/2.34.2",
+    "Accept-Encoding": "gzip, deflate",
+    "Accept": "*/*",
+    "Connection": "keep-alive",
+    "Authorization": "Bearer ",
+    "Content-Length": "725",
+    "Content-Type": "application/json"
+  },
+  "body_text": "{\"model\": \"glm-5.2-colibri\", \"prompt\": [[154822, 154824, 14572, 25, 55638, 614, 1012, 2952, 311, 5656, 20683, 504, 264, 34184, 1119, 279, 32616, 315, 264, 12722, 52938, 13, 3197, 279, 52938, 87894, 11, 432, 18611, 14065, 429, 5610, 264, 12827, 1730, 304, 34184, 80264, 13, 1096, 12827, 646, 1221, 387, 27341, 323, 1483, 311, 1281, 3746, 714, 29037, 7236, 369, 17409, 15768, 96457, 11, 77920, 11, 323, 14126, 13, 576, 2297, 1558, 537, 5240, 279, 52938, 894, 6646, 11, 714, 1045, 1251, 1744, 429, 432, 374, 537, 1290, 311, 912, 20683, 504, 279, 32616, 315, 825, 9859, 311, 2441, 13, 15908, 3409, 1850, 16539, 419, 4643, 5267, 16127, 25, 220, 6955]], \"temperature\": 0, \"max_tokens\": 1, \"logprobs\": 1, \"seed\": 1234, \"echo\": true}",
+  "body_json": {
+    "model": "glm-5.2-colibri",
+    "prompt": [
+      [
+        154822,
+        154824,
+        14572,
+        25,
+        55638,
+        614,
+        1012,
+        2952,
+        311,
+        5656,
+        20683,
+        504,
+        264,
+        34184,
+        1119,
+        279,
+        32616,
+        315,
+        264,
+        12722,
+        52938,
+        13,
+        3197,
+        279,
+        52938,
+        87894,
+        11,
+        432,
+        18611,
+        14065,
+        429,
+        5610,
+        264,
+        12827,
+        1730,
+        304,
+        34184,
+        80264,
+        13,
+        1096,
+        12827,
+        646,
+        1221,
+        387,
+        27341,
+        323,
+        1483,
+        311,
+        1281,
+        3746,
+        714,
+        29037,
+        7236,
+        369,
+        17409,
+        15768,
+        96457,
+        11,
+        77920,
+        11,
+        323,
+        14126,
+        13,
+        576,
+        2297,
+        1558,
+        537,
+        5240,
+        279,
+        52938,
+        894,
+        6646,
+        11,
+        714,
+        1045,
+        1251,
+        1744,
+        429,
+        432,
+        374,
+        537,
+        1290,
+        311,
+        912,
+        20683,
+        504,
+        279,
+        32616,
+        315,
+        825,
+        9859,
+        311,
+        2441,
+        13,
+        15908,
+        3409,
+        1850,
+        16539,
+        419,
+        4643,
+        5267,
+        16127,
+        25,
+        220,
+        6955
+      ]
+    ],
+    "temperature": 0,
+    "max_tokens": 1,
+    "logprobs": 1,
+    "seed": 1234,
+    "echo": true
+  }
+}

c/tests/golden_fixture_capture.py

@@ -22,8 +22,9 @@
     python3 tests/golden_fixture_capture.py diff fixtures_pre fixtures_post
 
 Battery: chat, chat+tools, chat streaming, completions without logprobs, and
-the error cases whose behavior must not move (seed 400, array-prompt 400,
-logprobs 400, out-of-range temperature 400) plus /v1/models.
+the error/no-op cases whose behavior must not move (seed accepted-and-ignored,
+an array `prompt` classified as one flat token-id prompt, logprobs served on
+the glm engine, out-of-range temperature 400) plus /v1/models.
 
 Normalization: every "id"/"created" field (recursively, and per SSE event) is
 replaced with a constant; nothing else is touched. Generation-bearing requests
@@ -76,11 +77,11 @@ def battery(model):
         ("completions_stop", "POST", "/v1/completions",
          {"model": model, "prompt": "Count: one, two,",
           "max_tokens": 16, "temperature": 0, "stop": ["five"]}),
-        ("err_seed", "POST", "/v1/completions",
+        ("seed_accepted", "POST", "/v1/completions",
          {"model": model, "prompt": "hello", "max_tokens": 1, "seed": 1234}),
-        ("err_array_prompt", "POST", "/v1/completions",
+        ("array_prompt_token_ids", "POST", "/v1/completions",
          {"model": model, "prompt": [1, 2, 3], "max_tokens": 1}),
-        ("err_logprobs", "POST", "/v1/completions",
+        ("logprobs_served", "POST", "/v1/completions",
          {"model": model, "prompt": "hello", "max_tokens": 1, "logprobs": 1}),
         ("err_bad_temperature", "POST", "/v1/chat/completions",
          {"model": model, "messages": chat_messages,

c/tests/test_anthropic_messages.py

@@ -34,6 +34,8 @@ def generate(self, prompt, maximum, temperature, top_p, on_text, cache_slot=0,
                  on_tool=None):
         self.prompts.append(prompt)
         self.emitted = 0
+        if on_accept is not None:  # simulate the engine's ACCEPT frame (#597)
+            on_accept({"prompt_tokens": 11})
         for chunk in self.script:
             on_text(chunk)
             self.emitted += 1