[Benchmark] Add GSM8K Platinum Eval (#14565)

benchmark/gsm8k/README.md

@@ -1,5 +1,15 @@
 ## Run benchmark
 
+### Using GSM8K Platinum
+
+GSM8K Platinum is a revised version of the GSM8K test set with corrected labels and removed ambiguous questions. It can be more stable than the original GSM8K dataset. It's a drop-in replacement that can be used by adding the `--platinum` flag:
+
+```
+python3 bench_sglang.py --num-shots 8 --num-questions 1209 --parallel 1209 --platinum
+```
+
+For more information, see: https://huggingface.co/datasets/madrylab/gsm8k-platinum
+
 ### Benchmark sglang
 ```
 python -m sglang.launch_server --model-path meta-llama/Llama-2-7b-chat-hf --port 30000

benchmark/gsm8k/bench_other.py

@@ -7,6 +7,7 @@ import time
 from concurrent.futures import ThreadPoolExecutor
 
 import numpy as np
+from datasets import load_dataset
 from tqdm import tqdm
 
 from sglang.test.test_utils import add_common_other_args_and_parse, get_call_generate
@@ -45,9 +46,16 @@ def main(args):
     call_generate = get_call_generate(args)
 
     # Read data
-    url = "https://raw.githubusercontent.com/openai/grade-school-math/master/grade_school_math/data/test.jsonl"
+    if args.platinum:
-    filename = download_and_cache_file(url)
+        print("Loading GSM8K Platinum dataset from HuggingFace...")
-    lines = list(read_jsonl(filename))
+        dataset = load_dataset("madrylab/gsm8k-platinum", "main", split="test")
+        lines = [
+            {"question": item["question"], "answer": item["answer"]} for item in dataset
+        ]
+    else:
+        url = "https://raw.githubusercontent.com/openai/grade-school-math/master/grade_school_math/data/test.jsonl"
+        filename = download_and_cache_file(url)
+        lines = list(read_jsonl(filename))
 
     # Construct prompts
     num_questions = args.num_questions
@@ -128,7 +136,7 @@ def main(args):
 
     with open(args.result_file, "a") as fout:
         value = {
-            "task": "gsm8k",
+            "task": "gsm8k-platinum" if args.platinum else "gsm8k",
             "backend": args.backend,
             "num_gpus": 1,
             "latency": round(latency, 3),
@@ -147,5 +155,10 @@ if __name__ == "__main__":
     parser.add_argument("--num-shots", type=int, default=5)
     parser.add_argument("--data-path", type=str, default="test.jsonl")
     parser.add_argument("--num-questions", type=int, default=200)
+    parser.add_argument(
+        "--platinum",
+        action="store_true",
+        help="Use GSM8K Platinum dataset (drop-in replacement with corrected labels)",
+    )
     args = add_common_other_args_and_parse(parser)
     main(args)

benchmark/gsm8k/bench_sglang.py

@@ -6,6 +6,7 @@ import re
 import time
 
 import numpy as np
+from datasets import load_dataset
 
 from sglang.lang.api import set_default_backend
 from sglang.test.test_utils import (
@@ -48,11 +49,18 @@ def main(args):
     set_default_backend(select_sglang_backend(args))
 
     # Read data
-    data_path = args.data_path
+    if args.platinum:
-    url = "https://raw.githubusercontent.com/openai/grade-school-math/master/grade_school_math/data/test.jsonl"
+        print("Loading GSM8K Platinum dataset from HuggingFace...")
-    if not os.path.isfile(data_path):
+        dataset = load_dataset("madrylab/gsm8k-platinum", "main", split="test")
-        data_path = download_and_cache_file(url)
+        lines = [
-    lines = list(read_jsonl(data_path))
+            {"question": item["question"], "answer": item["answer"]} for item in dataset
+        ]
+    else:
+        data_path = args.data_path
+        url = "https://raw.githubusercontent.com/openai/grade-school-math/master/grade_school_math/data/test.jsonl"
+        if not os.path.isfile(data_path):
+            data_path = download_and_cache_file(url)
+        lines = list(read_jsonl(data_path))
 
     # Construct prompts
     num_questions = args.num_questions
@@ -125,7 +133,7 @@ def main(args):
 
     with open(args.result_file, "a") as fout:
         value = {
-            "task": "gsm8k",
+            "task": "gsm8k-platinum" if args.platinum else "gsm8k",
             "backend": args.backend,
             "num_gpus": 1,
             "latency": round(latency, 3),
@@ -144,5 +152,10 @@ if __name__ == "__main__":
     parser.add_argument("--num-shots", type=int, default=5)
     parser.add_argument("--data-path", type=str, default="test.jsonl")
     parser.add_argument("--num-questions", type=int, default=200)
+    parser.add_argument(
+        "--platinum",
+        action="store_true",
+        help="Use GSM8K Platinum dataset (drop-in replacement with corrected labels)",
+    )
     args = add_common_sglang_args_and_parse(parser)
     main(args)