x

Signed-off-by: SumanthRH <[email protected]>
NovaSky-AI · Feb 6, 2025 · 7afb49f · 7afb49f
1 parent b9c0260
commit 7afb49f
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/skythought/skythought_evals/README.md b/skythought/skythought_evals/README.md
@@ -20,7 +20,7 @@ We provide a wrapper script `eval.py` to conveniently run reasoning benchmarks.
 **NOTE**: For reproducing `Sky-T1-32B-Preview` results on `AIME` and `GPQADiamond` dataset, pass in temperatures as `0.7`. 
 
 ```shell
-python -m skythought_evals.eval --model NovaSky-AI/Sky-T1-32B-Preview --evals=aime,gpqa_diamond --tp=8 --temperatures 0.7 --n 8
+python -m skythought_evals.eval --model NovaSky-AI/Sky-T1-32B-Preview --evals=aime,gpqa_diamond --tp=8 --temperatures 0.7 --n 16
 ```
 
 #### Example Usage

diff --git a/skythought/skythought_evals/inference_and_check.py b/skythought/skythought_evals/inference_and_check.py
@@ -666,7 +666,7 @@ def main():
     if args.result_dir and not os.path.exists(args.result_dir):
         os.makedirs(args.result_dir)
     temperature_str = ",".join(map(str, temperatures))
-    file_suffix = f"{model_config.name}_{args.task}_{args.split}_subset_{args.subset}_filter_{args.filter_difficulty}"
+    file_suffix = f"{model_config.name}_{args.task}_{args.split}_subset_{args.subset}_filter_{args.filter_difficulty}" + \
     f"_s{args.start}_e{args.end}_t{temperature_str}_n{args.n}"
     if (
         args.math_difficulty_lower_bound is not None