intel intel-extension-for-deepspeed issues

intel / intel-extension-for-deepspeed

Intel® Extension for DeepSpeed* is an extension to DeepSpeed that brings feature support with SYCL kernels on Intel GPU(XPU) device. Note XPU is already supported in stock DeepSpeed (upstream).

MIT License

56 stars 19 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

error for install deepspeed

#90 yash3056 opened 2 weeks ago
0
Add some huggingface transformers example

#89 Liangliang-Ma closed 1 month ago
2
[Zero2_Config]Increase the bucket_size to align with Zero3

#88 ys950902 closed 2 months ago
0
Add the llm inference script to run on borealis.

#87 ys950902 closed 2 months ago
0
[manifest] update mainfest to add h file in idex

#86 ys950902 closed 2 months ago
0
set comm_overlap as true by default for Zero3

#85 ys950902 closed 2 months ago
1
[XPU]Use host time to replace xpu time when IPEX version slower than 2.5

#84 ys950902 closed 2 months ago
1
[XPU] port xpu upstream back

#83 YizhouZ closed 2 months ago
0
Update README.md

#82 rogerxfeng8 closed 2 months ago
0
, in get_accelerator ds_accelerator = XPU_Accelerator() TypeError: Can't instantiate abstract class XPU_Accelerator with abstract methods

#81 uniartisan closed 2 months ago
1
sync with ipex.deepspeed

#80 baodii opened 4 months ago
0
[Config]modify the zero2 config align to zero3

#79 ys950902 closed 3 months ago
1
Set overlap_comm=false to use the same stream for computation and compuatation

#78 ys950902 closed 2 months ago
0
[Stream] Use the same stream for computation and communication

#77 ys950902 closed 4 months ago
0
fix quantization error

#76 baodii closed 4 months ago
0
[FLASH ATTN] fix fmha forward path acc issue

#75 YizhouZ closed 2 months ago
0
[config] set fused_adam optimizer by default.

#74 ys950902 closed 4 months ago
0
aligned with CUDA_Accelerator to fix moe issue

#73 shiyang-weng closed 2 months ago
0
[XPU Accelerator] add compile backend

#72 YizhouZ closed 5 months ago
0
add security.md

#71 rogerxfeng8 closed 5 months ago
0
align accelerator with Deepspeed

#70 Liangliang-Ma closed 5 months ago
0
[zero1+pp]Remove the config which is not needed

#69 ys950902 closed 4 months ago
0
add requirements.txt

#68 YizhouZ closed 5 months ago
0
[FLASH ATTN] Update flash-attn kernel implementation and compilation

#67 YizhouZ closed 6 months ago
0
update bf16 data type as upstream

#66 rogerxfeng8 closed 6 months ago
0
xpu_accelerator.py: add missing methods

#65 YizhouZ closed 7 months ago
0
Support bf16 type for transformer inference kernel to support Ds_Chat

#64 ys950902 closed 8 months ago
2
update woq builder and kernels; optimize context.h code getCurrentStream

#63 baodii closed 8 months ago
2
xpu_accelerator.py: add graph operations

#62 YizhouZ closed 9 months ago
0
xpu_accelerator.py: add abstract method export_envs

#61 YizhouZ closed 9 months ago
0
update all kernels in idex with dpct

#60 baodii closed 8 months ago
3
update idex dpct OpBuilders

#59 baodii closed 9 months ago
2
delete path added by mistake

#58 Liangliang-Ma closed 9 months ago
0
Baodi/support woq

#57 baodii closed 9 months ago
1
flash_attn.py: disable default debug build by removing "-g" flag

#56 YizhouZ closed 9 months ago
0
modify the generate_hostfile to support running on oam system

#55 ys950902 closed 9 months ago
1
[DeepSpeed-Chat]Add transform_inference sycl kernels to support hybrid-engine for DeepSpeed-Chat

#54 ys950902 closed 8 months ago
1
xpu_accelerator.py: return reasonable available memory

#53 YizhouZ closed 11 months ago
0
flash_attn: fix accuracy issue

#52 YizhouZ closed 11 months ago
1
add checkpoint flag

#51 YizhouZ closed 11 months ago
0
support available_memory

#50 YizhouZ closed 11 months ago
1
support flash_attn v2

#49 YizhouZ closed 11 months ago
3
fix build error with DeepSpeed triton support

#48 jinyouzhi closed 1 year ago
2
aio opbuilder in idex

#47 Liangliang-Ma closed 12 months ago
0
Can't instantiate abstract class XPU_Accelerator.

#46 teabagk7 closed 1 year ago
1
examples: allow adding shell para to run gpt

#45 Liangliang-Ma closed 1 year ago
0
add RoPE flag, don't merge

#44 Yejing-Lai opened 1 year ago
0
fix transformer inference kernels AOT build

#43 dc3671 closed 1 year ago
0
examples/generate_hostfile.sh: fix path error

#42 YizhouZ closed 1 year ago
0
examples: fix path error

#41 YizhouZ closed 1 year ago
0