Remove claims from code comments - keep purely factual

Author: Anivar A Aravind

language/deepseek-r1/PREPROCESSING.md

@@ -19,7 +19,7 @@ Let's think about this step by step.<|im_end|>
 
 ## Answer Extraction
 ```python
-# Remove reasoning, extract final answer
+# Extract final answer
 output = full_output.split('<|/thinking|>')[-1] if '<|/thinking|>' in full_output else full_output
 answer = re.search(r'####\s*(\d+)', output)
 final_answer = answer.group(1) if answer else output.strip()

language/llama3.1-8b/PREPROCESSING.md

@@ -2,7 +2,7 @@
 
 ## Tokenization
 ```python
-tokenizer = AutoTokenizer.from_pretrained(model_path)  # Use model's tokenizer
+tokenizer = AutoTokenizer.from_pretrained(model_path)
 max_length = 2048
 ```
 
@@ -20,6 +20,6 @@ Summarize this article:
 
 ## Verification
 ```bash
-# Check first token is 128000 (begin_of_text)
+# Verify first token
 python -c "print(tokenizer.encode('<|begin_of_text|>')[0])"
 ```