support min_p sampling parameter #2420
Merged
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
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 4, 2024
| @@ -23,27 +23,29 @@ | |||
|
|
|||
| namespace turbomind { | |||
|
|
|||
| struct DynamicDecodeCommonArgs { | |||
There was a problem hiding this comment.
Like struct AttentionParam, struct ModelParam, let's use struct DecodeParam
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 4, 2024
| template<typename T> | ||
| void invokeTopPSort(TopPSortParams& params, cudaStream_t stream); | ||
|
|
||
| struct TopPMinPFilterParams { |
There was a problem hiding this comment.
struct TopPMinPFilterParams is a subset of struct TopPSortPrams except the float* min_ps.
Can we merge them together?
lvhan028
reviewed
Sep 4, 2024
lvhan028
reviewed
Sep 5, 2024
lzhangzz
reviewed
Sep 6, 2024
lzhangzz
reviewed
Sep 6, 2024
| float thread_sum = s_sum; | ||
| topk_tmp_id_buf += batch_id * stride; | ||
| for (int i = tid; i < k; i += BLOCK_SIZE) { | ||
| sorted_logits[i] = s_val2[i] / thread_sum; |
There was a problem hiding this comment.
It seems that top-k doesn't use the 1e-6f eps like top-p does?
lvhan028
approved these changes
Sep 7, 2024
lzhangzz
approved these changes
Sep 9, 2024
|
Very excited to test this - thank you! 🙏 |
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Outdated suggestions cannot be applied.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.
support min_p sampling option
#1745