PyTPU: Migration of Python Code for Heterogenous Acceleration with Automated Test Generation
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
Software applications are increasingly built to take advantage of heterogeneous architectures as specialised hardware accelerators take centre stage in today’s computing environment. Tensor processing units (TPUs), the latest hardware addition, have demonstrated orders of magnitude improvement in computing efficiency over CPUs and GPUs for heterogeneous deep learning applications. However, despite the trend of incorporating heterogeneity and specialization in hardware, the creation of heterogeneous applications is confined to a handful of engineers. We propose a framework called PyTPU that takes Python code written in the PyTorch framework as input and automatically migrates it to its TPU compatible counterpart with test behaviour preservation and increased performance. First, PyTPU generates unit test cases to ensure test behavior compatibility. Second, using the abstract syntax tree and a manually curated exhaustive knowledge base, PyTPU migrates the original code to its TPU compatible version. Finally, PyTPU ensures the migrated code is readable and maintainable by adding necessary comments and conforming to PEP8 standard if needed. We evaluated PyTPU on four real-world Python applications with TPU v2 kernels. On average the migrated heterogeneous code is 19.103% faster than the original code while safeguarding test behavior preservation.
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 | 0.000 | 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 | 0.000 |
| Open science | 0.000 | 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