[ Docs ] Conceptual Guides #18
Open
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
robertgshaw2-neuralmagic
changed the title
robertgshaw2-neuralmagic
changed the title
bfineran
reviewed
Jul 15, 2024
|
|
||
| ## Theory | ||
|
|
||
| Performing quantization to go from `float16` to `int8` (or lower) is tricky. Only 256 values can be represented in `int8`, while `float16` can represent a very wide range of values. The idea is to find the best way to project our range [a, b] of `float32` values to the `int8` space. |
There was a problem hiding this comment.
good space for a diagram (bucketing the weight distribution to 256 buckets)
|
|
||
| * **Static quantization**: the range for each activation is computed in advance at quantization-time, typically by passing representative "calibration" data through the model and recording the activation values. In practice, we run a number of forward passes on a calibration dataset is done and compute the ranges according to the observed calibration data. | ||
|
|
||
| In general, it is best practice to start your experiments with: |
There was a problem hiding this comment.
Why is it best practice?
Co-authored-by: Benjamin Fineran <bfineran@users.noreply.github.com>
Co-authored-by: Benjamin Fineran <bfineran@users.noreply.github.com>
Co-authored-by: Benjamin Fineran <bfineran@users.noreply.github.com>
Co-authored-by: Benjamin Fineran <bfineran@users.noreply.github.com>
Co-authored-by: Benjamin Fineran <bfineran@users.noreply.github.com>
markmc
pushed a commit
to markmc/llm-compressor
that referenced
this pull request
Nov 13, 2024
* test forward (vllm-project#16) * test frozen (vllm-project#17) * test frozen * rename * lifecycle conftest (vllm-project#21) * test initalize (vllm-project#18) * test initalize * newline * parametrize weights and inp_act * remove dup * test lifecycle (vllm-project#19) * test lifecycle * comments * comments * add quantization test * Lifecycle/min max obs (vllm-project#20) * min max test * add minmax obs * test scale range and min_max update * rebase * rebase * fix * fix
# for free
to join this conversation on GitHub.
Already have an account?
# 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.
SUMMARY:
TEST PLAN: