Jason Liang

icon indicating a website link https://www.linkedin.com/in/jason-liang-3189b249

icon company Cognizant icon location San Francisco icon location Research Scientist, PhD from UT Austin Interests: Evolutionary Algorithms, AGI, LLMs

Results 3 issues of Jason Liang

How do initialize ddpg with a pretrained policy.

1 comment

Hi, If I have a policy (a tensorflow model) that has been trained using some other RL method, is there way for me to initalize ddpg with this policy?

Extracting hidden state of seq2seq model?

3 comment

Given a trained Seq2Seq (with 1 encoder and 1 decoder layer) model and an input sequence, how can I get the hidden state/vector of the encoder layer immediately after feeding...

Humaneval, use base model or instruct finetuned model?

1 comment

Which one would you recommend to use to get the best possible performance?