Hardware-aware Gradient-based Column-wise Mixed-Precision Quantization for Compute-in-Memory Accelerators
Bibliographic record
Abstract
Efficient deep learning deployment, especially on the Edge, requires quantization strategies that not only preserve model accuracy but also align with hardware constraints to maximize performance and efficiency. In this work, we propose an end-to-end framework for mixed-precision quantization of convolutional neural networks (CNNs), optimized for deployment on compute-in-memory (CIM) accelerators. Our approach employs column-wise mixed-precision quantization to balance model accuracy and hardware complexity. We implement the framework using fully gradient-based algorithms, enabling efficient design space exploration. To account for hardware constraints, we integrate an analog CIM accelerator simulator into the pipeline, providing hardware metrics. Additionally, an estimator network is co-trained with the task model to supply hardware-induced gradients, eliminating the need for equation-based intermediate approximations. We validate our framework on the CIFAR-10 dataset using VGG8, VGG11, and ResNet18, demonstrating its effectiveness in optimizing both model and hardware performance achieving 50.34 TOPs/W of energy efficiency, 0.15 TOPs/mm2of compute efficiency and 189.25 mm2of area with ResNet18. Further experiments on ImageNet dataset prove its cross-dataset adaptability and confirm its ability to direct control the hardware metrics.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.001 |
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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".