Create Example Inferencing Script.py

Example Inferencing Script.py

@@ -0,0 +1,61 @@
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import os
+
+def load_model_and_tokenizer(model_path):
+    # First, try loading from the directory
+    try:
+        print(f"Attempting to load model from directory: {model_path}")
+        model = AutoModelForCausalLM.from_pretrained(model_path)
+    except Exception as e:
+        print(f"Failed to load from directory. Error: {e}")
+        # If that fails, try loading the specific .safetensors file
+        safetensors_path = os.path.join(model_path, "model.safetensors")
+        if os.path.exists(safetensors_path):
+            print(f"Attempting to load model from file: {safetensors_path}")
+            model = AutoModelForCausalLM.from_pretrained(safetensors_path)
+        else:
+            raise ValueError(f"Could not find model at {model_path} or {safetensors_path}")
+    
+    # Load the tokenizer from the original DistilGPT2 model
+    tokenizer = AutoTokenizer.from_pretrained("distilgpt2")
+    
+    return model, tokenizer
+
+def generate_text(model, tokenizer, prompt, max_length=125, num_return_sequences=1):
+    input_ids = tokenizer.encode(prompt, return_tensors='pt')
+    
+    # Generate text
+    output = model.generate(
+        input_ids,
+        max_length=max_length,
+        num_return_sequences=num_return_sequences,
+        no_repeat_ngram_size=6,
+        top_k=25,
+        top_p=0.99,
+        temperature=0.34
+    )
+    
+    return [tokenizer.decode(seq, skip_special_tokens=True) for seq in output]
+
+def main():
+    model_path = r"literalpathtothefoldernamed\checkpoint-4000" #change this to where you have the folder on your computer
+    
+    print(f"Attempting to load model...")
+    model, tokenizer = load_model_and_tokenizer(model_path)
+    
+    print("Model loaded successfully. Enter prompts to generate text. Type 'quit' to exit.")
+    
+    while True:
+        prompt = input("Enter a prompt: ")
+        if prompt.lower() == 'quit':
+            break
+        
+        generated_texts = generate_text(model, tokenizer, prompt)
+        
+        print("\nGenerated Text:")
+        for i, text in enumerate(generated_texts, 1):
+            print(f"{i}. {text}\n")
+
+if __name__ == "__main__":
+    main()