The parameter accurateHalf can be set to true when creating a CPU or IPUModel target, in which case half will be correctly implemented as 16-bitIEEE floating point. This will be slower, but will produce the same results asthe IPU.
The parameter accurateHalf can be set to true when creating a CPU or IPUModel target, in which case half will be correctly implemented as 16-bitIEEE floating point. This will be slower, but will produce the same results asthe IPU.