5 papers
dMX: Differentiable Mixed-Precision Assignment for Low-Precision Floating-Point Formats
Giuseppe Franco, Ian Colbert, Pablo Monteagudo-Lago +2
Quantizing large language models (LLMs) to low-precision floating-point representations is central to efficient deployment, yet applying a single bit-width uniformly across all lay…
Signed Symmetric Quantization for Few-Bit Integers
Ian Colbert, Eashan Dash, Pablo Monteagudo-Lago +5
The signed integer alphabet contains one more negative representable value than positive. Yet, by convention, the standard symmetric integer quantizer fixes its scale to be strictl…
Optimal Post-Training Quantization Scales and Where to Find Them
Juan Amboage, Pablo Monteagudo-Lago, Ian Colbert +2
Post-training quantization (PTQ) compresses large language models by mapping weights to low-bit representations. The scaling factor that defines the quantization grid is typically…
Improving Quantization with Post-Training Model Expansion
Giuseppe Franco, Pablo Monteagudo-Lago, Ian Colbert +2
The size of a model has been a strong predictor of its quality, as well as its cost. As such, the trade-off between model cost and quality has been well-studied. Post-training opti…
Accumulator-Aware Post-Training Quantization for Large Language Models
Ian Colbert, Giuseppe Franco, Fabian Grob +2
When quantizing weights and activations to increasingly narrower representations, the cost of additions begins to dominate that of multiplications in multiply-accumulate (MAC) unit…