diff --git a/.github/workflows/benchmark.yml b/.github/workflows/benchmark.yml
index 0dd5cca0f8..dcf555097b 100644
--- a/.github/workflows/benchmark.yml
+++ b/.github/workflows/benchmark.yml
@@ -81,6 +81,8 @@ jobs:
       run: |
         CUDA=1 JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
         CUDA=1 JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
+    - name: Run GPT2 w HALF/BEAM
+      run: CUDA=1 JIT=1 HALF=1 BEAM=4 CACHELEVEL=0 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half_beam.txt
     - uses: actions/upload-artifact@v3
       with:
         name: Speed (NVIDIA)
@@ -89,6 +91,7 @@ jobs:
           torch_speed.txt
           gpt2_unjitted.txt
           gpt2_jitted.txt
+          gpt2_half_beam.txt
 
   testamdbenchmark:
     name: AMD Benchmark
diff --git a/examples/gpt2.py b/examples/gpt2.py
index 7306d5d7a8..108a14134a 100644
--- a/examples/gpt2.py
+++ b/examples/gpt2.py
@@ -159,9 +159,10 @@ class GPT2:
 if __name__ == "__main__":
   Tensor.no_grad = True
   print(f"using {Device.DEFAULT} backend")
+  default_prompt = "What is the answer to life, the universe, and everything?"
 
   parser = argparse.ArgumentParser(description='Run GPT2 in tinygrad', formatter_class=argparse.ArgumentDefaultsHelpFormatter)
-  parser.add_argument('--prompt', type=str, default="What is the answer to life, the universe, and everything?", help="Phrase to start with")
+  parser.add_argument('--prompt', type=str, default=default_prompt, help="Phrase to start with")
   parser.add_argument('--count', type=int, default=100, help="Max number of tokens to generate")
   parser.add_argument('--temperature', type=float, default=0.8, help="Temperature in the softmax")
   parser.add_argument('--model_size', type=str, default="gpt2-medium", help="Size of model to use [gpt2, gpt2-medium, gpt2-large, gpt2-xl]")
@@ -191,4 +192,16 @@ if __name__ == "__main__":
       print('Generating text...')
       if len(texts) == 1: print(texts[0])
       else:
-        for i,text in enumerate(texts): print(colored(f"Response {i}:", "green"), text)
\ No newline at end of file
+        for i,text in enumerate(texts): print(colored(f"Response {i}:", "green"), text)
+
+    # validate output!
+    if args.temperature == 0 and args.model_size == "gpt2-medium" and args.count == 10:
+      expected = {
+        default_prompt: "What is the answer to life, the universe, and everything?\n\nThe answer is that we are all one",
+        "Hello.": "Hello. I'm a little late to the party, but",
+      }
+      try:
+        assert texts[0] == expected[args.prompt]
+        print(colored("output validated", "green"))
+      except KeyError:
+        pass