Automatic Optimize-time Validation for Binary Optimizers
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
We propose an approach called automatic optimize-time validation for binary optimizers. Our approach does not involve executing the whole program for validation but selecting a small part of code (1 to 100 instructions) for the target test code. It executes the target code and its optimized code with several input data during binary optimization. One benefit is that we can test a small part of an actual customer's code during binary optimization. Our approach can be used to test several input data not included in the target code, which is the most beneficial aspect of the approach. If the results are the same after execution, we will use the optimized code for the final output code. If the results differ, we can consider a couple of option, e.g., while developing a binary optimizer, we can abort the compilation with an error message to easily detect a bug. After a binary optimizer becomes generally available, we can use the input code for the final output code to maintain compatibility. Our goal is for the output binary code to be compatible, fast, and small. We focused on how to improve compatibility in this study. We implemented our approach in our binary optimizer and successfully detected one new bug. We used a very small binary program to observe the worst case of increased compilation time and output binary file size. Our implementation showed that our approach increases optimization time by only 0.02% and output binary file size by 8%.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.004 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it