Use power-of-two scaling in autoscale scaled translation ops rules. #65
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
balancap
force-pushed
the
use-power-of-two-scaling-factors-in-unit-scaling-rules
branch
from
a8d2bdd
to
da22330
Compare
balancap
changed the title
autoscale ops rules
balancap
force-pushed
the
use-power-of-two-scaling-factors-in-unit-scaling-rules
branch
from
da22330
to
eede38a
Compare
balancap
force-pushed
the
use-power-of-two-scaling-factors-in-unit-scaling-rules
branch
4 times, most recently
from
df52131
to
8971b50
Compare
As shown in #60 issue, propagating non power-of-two scaling factors can decrease training accuracy in low precision (typically in FP16). The additional rescaling operations will introduce non-negligible floating point accumulated errors. Ths PR is adding the option to round the scale to a power-of-two in scaled translation. Supporting at the moment only rounding up and down. The rounding mode can be modified in the config dataclass `AutoScaleConfig`. Scaled translations updated are: `dot_general`, `add`, `sub` and `reduce_sum`. Finally, when implicitely converting scalars to scaled arrays, the method `make_scaled_scaled` now splits the input mantissa and exponent between data and scale.
balancap
force-pushed
the
use-power-of-two-scaling-factors-in-unit-scaling-rules
branch
from
8971b50
to
81b876c
Compare
balancap
deleted the
use-power-of-two-scaling-factors-in-unit-scaling-rules
branch
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As shown in #60 issue, propagating non power-of-two scaling factors can decrease training accuracy in low precision (typically in FP16).
The additional rescaling operations will introduce non-negligible floating point accumulated errors.
Ths PR is adding the option to round the scale to a power-of-two in scaled translation. Supporting at the moment only rounding up and down. The rounding mode can be modified in the config dataclass
AutoScaleConfig.Scaled translations updated are:
dot_general,add,subandreduce_sum.Finally, when implicitely converting scalars to scaled arrays, the method
make_scaled_scalednow splits the input mantissa and exponent between data and scale.