Tim Dettmers

Assistant Professor@Carnegie Mellon University·美国·OpenReview

研究方向

deep learning · quantization · agent systems · agents · model efficiency · language modeling · knowledge graph embeddings · parallelization of neural networks

8.7

SpQR: A Sparse-Quantized Representation for Near-Lossless LLM Weight Compression

ICLR 2024Poster

一作

6.3

Scaling Retrieval-Based Language Models with a Trillion-Token Datastore

NeurIPS 2024Poster

6.0

MatFormer: Nested Transformer for Elastic Inference

合作者 (20)

Tim Dettmers

OLMoE: Open Mixture-of-Experts Language Models

Holistically Evaluating the Environmental Impact of Creating Language Models

Seeded LoRA: Collaborative Fine-Tuning Through Seed Initialization of Adapters

MatFormer: Nested Transformer for Elastic Inference

SpQR: A Sparse-Quantized Representation for Near-Lossless LLM Weight Compression

Scaling Retrieval-Based Language Models with a Trillion-Token Datastore

MatFormer: Nested Transformer for Elastic Inference