What methods are used to increase token throughput of LLM inference without sacrificing ability?
Learning Technology Interview Questions
2,887 learning technology interview questions shared by candidates
try to implement logistic regression
Explain PPO, loss function of DPO
Implement Single head and Multi head attention in pytorch. A follow up to this was use masking to mask out the later-on tokens. Also asked me to implement KL divergence in python (but maybe this was because I talked about kl-divergence in my projects)
30 mins ML questions regarding the resume, 30 mins LC medium (BFS)
about language models, if know any new LLMs, know transformer, know NLP, know CV, and know about knowledge graph.
Why applying TikTok? Past project deep dive.
What previous ML experience do you have
What is the difference between supervised and unsupervised learning. Give examples.
What does the vocabulary space of a Language Model mean?
Viewing 2131 - 2140 interview questions