Search for: All records

We propose SLOPE, a Double-Pruned Sparse Plus Lazy Low-rank Adapter Pretraining method for LLMs that improves the accuracy of sparse LLMs while accelerating their pretraining and inference and reducing their memory footprint. Sparse pretraining of LLMs reduces the accuracy of the model, to overcome this, prior work uses dense models during fine-tuning. SLOPE improves the accuracy of sparsely pretrained models by adding low-rank adapters in the final 1% iterations of pretraining without adding significant overheads to the model pretraining and inference. In addition, SLOPE uses a double-pruned backward pass formulation that prunes the transposed weight matrix using N:M sparsity structures to enable an accelerated sparse backward pass. SLOPE accelerates the training and inference of models with billions of parameters up to 1.25→ and 1.54→ respectively (OPT-33B and OPT-66B) while reducing their memory usage by up to 0.63→ and 0.61→ for training and inference respectively. 

Recycling glass fiber reinforced polymer (GFRP) composite materials has been proven to be challenging due to their high mechanical performance and high resistance to harsh chemical and thermal conditions. This work discusses the efforts made in the past to mechanically process GFRP waste materials by cutting them into large-sized (cm scale) pieces, as opposed to pulverization, for use in concrete mixtures. These pieces can be classified into two main categories—coarse aggregate and discrete reinforcement, here referred to as “needles.” The results from all the studies show that using GFRP coarse aggregate leads to significant reductions in the compressive strength and tensile strength of concrete. However, GFRP needles lead to sizable increases in the energy absorption capacity of concrete. In addition, if the glass fibers are longitudinally aligned within the needles, these elements can substantially increase the tensile strength of concrete. Processing GFRP waste into needles requires less energy and time than that for producing GFRP coarse aggregate. Also, compared to pulverized GFRP waste, which consists of broken and separate particles of glass and resin that at best can be used as low-quality fillers, GFRP needles are high strength composite elements