"symptom","description"
"None","The problem does not cause any symptoms."
"Poor performance","There is performance degradation"
"Missing variable value","Variables are undefined"
"Exception handling error","There was an error in exception handling"
"No support for eager mode","The function was not supported in eager mode."
"Dynamic features unavailable","By using tf.function, they are making Python into a compiled language. The code would have been fine (due to lexical scoping) to run dynamically but because it is now compiled by Autograph, we have a problem."
"Autograph warning","The developer get an Autograph warning and it could not transform the function and runs it as-is."
"Unknown","Unable to identify the problem symptom."
"Incorrect learning result","Either in classification or prediction"
"Runtime error","An uncaught thrown exception. For example, a missing field."
"Unexpected behavior","The tf.function block produces an unexpected result/output."
"Failing test","We are getting assertion errors"
"Poor debuggability","Debugging code was made difficult."
"Hang","Occurs when either a process or system ceases to respond to inputs"
"Retracing warnings","Excessive retracing warnings which means it may be creating unnecessary graphs"