Repository navigation

All pull requests

Labels (9)Milestones (0)New pull request

Open

(38)

Closed

(131)

fix: fix position_encoding_table memory error.
#791johnson-magic opened Mar 27, 2024
Fix shape mismatch on the masked_tokens param in decoder masked multi-head attention kernel.
#773FengDSP opened Oct 24, 2023
Include stdio.h
#770JihaoXin opened Oct 19, 2023
2
Ft llama opt
#762dypshong opened Oct 2, 2023
Support Seq length up to 8K
#756zhen-jia opened Sep 4, 2023
[BugFix] GPT inference error when pipeline_para_size > 1 and int8_mode != 0
#75000why00 opened Aug 23, 2023
[Bugfix] GptJ & GptNeoX batch inference error
#742YZP17121579 opened Aug 11, 2023
1
Add fusion-for-decoder-only for llama
#733binxuan opened Jul 28, 2023
Fix beam search output_log_prob index error
#732cpm0722 opened Jul 25, 2023
[Doc] Add projects section in README which is developed based on FasterTransformer
#731lvhan028 opened Jul 25, 2023
2
Add triton fastertransformer backend support for deberta
#725sfc-gh-zhwang opened Jul 19, 2023
1
Add cuDNN include path as a common include dir
#724jacobkahn opened Jul 18, 2023
fix: initialize tiled_prompt_lengths_buf_ to zero in gptneox
#716yandai opened Jul 13, 2023
Remove parenthesis from asserts
#699miguelusque opened Jul 2, 2023
Huggingface gptj convert script supports sharded checkpoint
#695skyser2003 opened Jun 29, 2023
[Doc] Fix typo in gpt_guide.md
#682myry96 opened Jun 26, 2023
swin-transformer quantization readme files changes
#675Mhhhaster opened Jun 16, 2023
fix: fix Qk_vec_acum_fp32_ has already been declared
#659lkm2835 opened Jun 9, 2023
gptneox & gptj int8 quantization & share context
#653rahuan opened Jun 7, 2023
Add missing headers
#648brian14708 opened Jun 1, 2023
Fix TOC of gptneox_guide.md
#633xu-song opened May 23, 2023
Update gpt_guide.md: documentation link is invalid
#620treycheng opened May 22, 2023
fix multi-gpu build
#616dskhudia opened May 17, 2023Contributor
1
Fix mpi library linking issue
#612liangfu opened May 16, 2023
remove buggy duplicated code
#609chenho74 opened May 12, 2023

Navigation Menu

Repository navigation

All pull requests

fix: fix position_encoding_table memory error.

Fix shape mismatch on the masked_tokens param in decoder masked multi-head attention kernel.

Include stdio.h

Ft llama opt

Support Seq length up to 8K

[BugFix] GPT inference error when pipeline_para_size > 1 and int8_mode != 0

[Bugfix] GptJ & GptNeoX batch inference error

Add fusion-for-decoder-only for llama

Fix beam search output_log_prob index error

[Doc] Add `projects` section in README which is developed based on FasterTransformer

Add triton fastertransformer backend support for deberta

Add cuDNN include path as a common include dir

fix: initialize tiled_prompt_lengths_buf_ to zero in gptneox

Remove parenthesis from asserts

Huggingface gptj convert script supports sharded checkpoint

[Doc] Fix typo in gpt_guide.md

swin-transformer quantization readme files changes

fix: fix Qk_vec_acum_fp32_ has already been declared

gptneox & gptj int8 quantization & share context

Add missing headers

Fix TOC of gptneox_guide.md

Update gpt_guide.md: documentation link is invalid

fix multi-gpu build

Fix mpi library linking issue

remove buggy duplicated code

Pull requests