Spaces:
Running
Running
Danny Liu commited on
Commit ·
14a2644
1
Parent(s): 9a205f0
fix weird bug about web content
Browse files- src/about.py +1 -1
src/about.py
CHANGED
|
@@ -26,7 +26,7 @@ NUM_FEWSHOT = 0 # Change with your few shot
|
|
| 26 |
# Your leaderboard name
|
| 27 |
TITLE = """<h1 align="center" id="space-title">How LLMs Fail and Generalize in RTL Coding for Hardware Design?</h1>"""
|
| 28 |
|
| 29 |
-
CONCLUSION_TEXT = """
|
| 30 |
Evaluations on the VerilogEval Human benchmark reveal a strict empirical ceiling, with frontier models plateauing at a 90.8% initial pass rate.
|
| 31 |
The solvability taxonomy exposes that L3U (Unsolvable) errors dominate across all model families, revealing persistent knowledge gaps that inference-time scaling cannot address.
|
| 32 |
Our analysis exposes a striking surface convergence gap: optimization drastically reduces syntax errors but concurrently increases functional testbench failures.
|
|
|
|
| 26 |
# Your leaderboard name
|
| 27 |
TITLE = """<h1 align="center" id="space-title">How LLMs Fail and Generalize in RTL Coding for Hardware Design?</h1>"""
|
| 28 |
|
| 29 |
+
CONCLUSION_TEXT = f"""
|
| 30 |
Evaluations on the VerilogEval Human benchmark reveal a strict empirical ceiling, with frontier models plateauing at a 90.8% initial pass rate.
|
| 31 |
The solvability taxonomy exposes that L3U (Unsolvable) errors dominate across all model families, revealing persistent knowledge gaps that inference-time scaling cannot address.
|
| 32 |
Our analysis exposes a striking surface convergence gap: optimization drastically reduces syntax errors but concurrently increases functional testbench failures.
|