Skip to content

Pull requests: NVIDIA/FasterTransformer

New pull request New
Clear current search query, filters, and sorts
40 Open 129 Closed
40 Open 129 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Remove parenthesis from asserts
#699 opened Jul 2, 2023 by miguelusque Loading…
fix: initialize tiled_prompt_lengths_buf_ to zero in gptneox
#716 opened Jul 13, 2023 by yandai Loading…
Add cuDNN include path as a common include dir
#724 opened Jul 18, 2023 by jacobkahn Loading…
Add triton fastertransformer backend support for deberta
#725 opened Jul 19, 2023 by sfc-gh-zhwang Loading…
1
[Doc] Add projects section in README which is developed based on FasterTransformer
#731 opened Jul 25, 2023 by lvhan028 Loading…
2
Fix beam search output_log_prob index error
#732 opened Jul 25, 2023 by cpm0722 Loading…
Add fusion-for-decoder-only for llama
#733 opened Jul 28, 2023 by binxuan Loading…
[Bugfix] GptJ & GptNeoX batch inference error
#742 opened Aug 11, 2023 by YZP17121579 Loading…
1
[BugFix] GPT inference error when pipeline_para_size > 1 and int8_mode != 0
#750 opened Aug 23, 2023 by 00why00 Loading…
Support Seq length up to 8K
#756 opened Sep 4, 2023 by zhen-jia Loading…
Ft llama opt
#762 opened Oct 2, 2023 by dypshong Loading…
Include stdio.h
#770 opened Oct 19, 2023 by JihaoXin Loading…
2
Fix shape mismatch on the masked_tokens param in decoder masked multi-head attention kernel.
#773 opened Oct 24, 2023 by FengDSP Loading…
Update README.md
#776 opened Oct 29, 2023 by eltociear Loading…
fix: fix position_encoding_table memory error.
#791 opened Mar 27, 2024 by johnson-magic Loading…
1
ProTip! Follow long discussions with comments:>50.