Top Questions

1. tf.tape.gradient() returns None for certain losses
   8 votes

   1 answers
2. No module named 'numpy' during docker build
   5 votes

   2 answers
3. tensorflow 2.0: tf.GradientTape().gradient() returns None
   2 votes

   1 answers
4. How is model parallelism implemented for GPT2 in MegatronLM?
   2 votes

   1 answers