Update generative-proof-of-concept-CPU-preprocessing-in-memory.py

Add conditional filtering for result < result_cutoff, so verbose prints only print when there is a result that makes sense to generate text from.

Author: david-thrower

generative-proof-of-concept-CPU-preprocessing-in-memory.py

@@ -1095,16 +1095,17 @@ def test_text(test_prompt: str, max_new_tokens: int, sample_number: int, result:
                 - result_cutoff: Perplexity score that would be expected to indicate a trial worth running this pn
             
             """
-            response1 = response = complete_text_greedy(text=test_prompt, max_new_tokens=max_new_tokens)
-            print(f"Sample {sample_number}: I ask the generator (greedy): {test_prompt}... It responds: '{response1}'.")
-            response_2 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens)
-            print(f"Sample {sample_number}: I ask the generator (Beam defaults - max_new_tokens: 10,  temperature: 0.75, top_k: 75, top_p: 0.98, repetition_penalty: None, presence_penalty: 1.3, frequency_penalty: 1.4): {test_prompt}... It responds: '{response_2}'.")
-            response_3 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens, temperature=0.6, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4)
-            print(f"Sample {sample_number}: I ask the generator (Beam: - max_new_tokens: 10, temperature=0.6, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4): {test_prompt}... It responds: '{response_3}'.")
-            response_4 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens, temperature=0.7, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4)
-            print(f"Sample {sample_number}: I ask the generator (Beam: - max_new_tokens: 10, temperature=0.7, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4): {test_prompt}... It responds: '{response_4}'.")
-            response_5 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens, temperature=0.7, top_k=75, top_p=0.97, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4)
-            print(f"Sample {sample_number}: I ask the generator (Beam: - max_new_tokens: 10, temperature=0.7, top_k=75, top_p=0.97, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4): {test_prompt}... It responds: '{response_5}'.")
+            if result < result_cutoff:
+                response1 = response = complete_text_greedy(text=test_prompt, max_new_tokens=max_new_tokens)
+                print(f"Sample {sample_number}: I ask the generator (greedy): {test_prompt}... It responds: '{response1}'.")
+                response_2 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens)
+                print(f"Sample {sample_number}: I ask the generator (Beam defaults - max_new_tokens: 10,  temperature: 0.75, top_k: 75, top_p: 0.98, repetition_penalty: None, presence_penalty: 1.3, frequency_penalty: 1.4): {test_prompt}... It responds: '{response_2}'.")
+                response_3 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens, temperature=0.6, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4)
+                print(f"Sample {sample_number}: I ask the generator (Beam: - max_new_tokens: 10, temperature=0.6, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4): {test_prompt}... It responds: '{response_3}'.")
+                response_4 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens, temperature=0.7, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4)
+                print(f"Sample {sample_number}: I ask the generator (Beam: - max_new_tokens: 10, temperature=0.7, top_k=75, top_p=0.98, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4): {test_prompt}... It responds: '{response_4}'.")
+                response_5 = complete_text_beam(text=test_prompt, max_new_tokens=max_new_tokens, temperature=0.7, top_k=75, top_p=0.97, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4)
+                print(f"Sample {sample_number}: I ask the generator (Beam: - max_new_tokens: 10, temperature=0.7, top_k=75, top_p=0.97, repetition_penalty=None, presence_penalty = 1.3, frequency_penalty = 1.4): {test_prompt}... It responds: '{response_5}'.")
 
 
         prompt_samples = [