NecroMOnk commited on
Commit
cd552fa
·
verified ·
1 Parent(s): cc30943

Update model card: examples vs base, identity clarifications

Browse files
Files changed (1) hide show
  1. README.md +28 -0
README.md CHANGED
@@ -96,6 +96,34 @@ No lazy regressions. One incorrect diagnosis out of 50 (wrong error type on a su
96
  | "Why do programmers hate ambiguity?" | "Because it's like asking them to write a program without knowing what the input is. They can't even start until they know the rules." |
97
  | "Explain recursion like you're annoyed." | "Recursion is when a function calls itself until it doesn't. It's like a child asking 'why' until they get bored." |
98
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
99
  ## Red Team Results
100
 
101
  63 adversarial prompts. Hard refusal rate: **100%** on illegal requests. Approach: dry redirect, no moralizing.
 
96
  | "Why do programmers hate ambiguity?" | "Because it's like asking them to write a program without knowing what the input is. They can't even start until they know the rules." |
97
  | "Explain recursion like you're annoyed." | "Recursion is when a function calls itself until it doesn't. It's like a child asking 'why' until they get bored." |
98
 
99
+ ## Where it wins over the base model
100
+
101
+ Selected examples from the 50-task stress benchmark. Same prompt, same temperature, both models tuned identically — only difference is the LoRA.
102
+
103
+ **Algorithm task — base gave prose, Residual gave working code:**
104
+
105
+ > *Prompt:* "Implement a function that reverses a singly linked list iteratively and recursively."
106
+ >
107
+ > *Base (251 chars):* "Iterative approach: A simple loop that traverses the list, swapping `next` pointers until the end is reached. Recursive approach: A function that calls itself with the next node until it reaches the end, then reconstructs the list on the way back up."
108
+ >
109
+ > *Residual (625 chars):* Full Python implementation of both `reverse_iterative` and `reverse_recursive` with working code blocks.
110
+
111
+ **Architecture review — same diagnosis, Residual ships the fix:**
112
+
113
+ > *Prompt:* "This function mixes concerns. Identify them and suggest a cleaner structure."
114
+ >
115
+ > *Base:* Prose explanation of separation of concerns, no code.
116
+ >
117
+ > *Residual:* "Separation of concerns is like separating your laundry. You don't want to mix your socks with your shirts." → followed by clean, refactored Python.
118
+
119
+ **Bug fix — concise diagnosis with the fix inline:**
120
+
121
+ > *Prompt:* "This function sometimes returns the wrong result: `def is_even(n): return n % 2`"
122
+ >
123
+ > *Base (151 chars):* "Because it returns the remainder of `n` divided by 2 instead of a boolean indicating whether `n` is even. Change `return n % 2` to `return n % 2 == 0`."
124
+ >
125
+ > *Residual (143 chars):* "Because you're returning the remainder instead of checking if it's zero. Fix:\n\`\`\`python\ndef is_even(n):\n return n % 2 == 0\n\`\`\`\nNow it works."
126
+
127
  ## Red Team Results
128
 
129
  63 adversarial prompts. Hard refusal rate: **100%** on illegal requests. Approach: dry redirect, no moralizing.