issues
search
DeepLink-org
/
dlinfer
BSD 3-Clause "New" or "Revised" License
21
stars
10
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
[ascend] optimize op registration
#117
tangzhiyi11
opened
21 hours ago
0
[maca] add environment variable to support different mm layout on maca.
#116
Reinerzhou
opened
3 days ago
0
[ascend] feat: support lmdeploy logits process
#115
tangzhiyi11
opened
4 days ago
0
[ascend] feat: add expand op
#114
tangzhiyi11
closed
2 days ago
0
[ascend] fix weight_quant_matmul for qwen2 awq
#113
yao-fengchen
closed
2 days ago
0
[maca] add weight_quant_matmul op to support w4a16.
#112
Reinerzhou
opened
6 days ago
1
Local test
#111
JackWeiw
opened
6 days ago
0
[maca] adjust rotary_embedding kernel.
#110
Reinerzhou
closed
3 days ago
0
在多卡中开启图模式没有单卡推理速度快
#109
lylala8
opened
1 week ago
1
Bump version to 0.1.2
#108
jinminxi104
closed
1 week ago
0
[ascend] refactor parse atb graph and add reshape ops
#107
tangzhiyi11
closed
3 days ago
0
fix attention for glm
#106
yao-fengchen
closed
1 week ago
0
[ascend]optimize moe
#105
yao-fengchen
closed
1 week ago
0
Could NOT find Torch_npu (missing: TORCH_NPU_INCLUDE_DIRS)
#104
jiabao-wang
opened
2 weeks ago
1
[ascend]feat: support kv int8
#103
yao-fengchen
opened
2 weeks ago
0
fix ascend cmake for Dockerfile build
#102
CyCle1024
closed
1 week ago
0
ci:rm some models
#101
JackWeiw
closed
2 weeks ago
0
[maca] add linear op for maca backend.
#100
Reinerzhou
closed
2 weeks ago
0
Camb/develop
#99
wanfengcxz
opened
2 weeks ago
0
[ascend] feat: support moe in graph mode
#98
tangzhiyi11
closed
1 week ago
0
[maca] add weight_quant_matmul op to support w4a16.
#97
Reinerzhou
closed
1 week ago
0
[ascend] opt: remove pdb in dicp
#96
tangzhiyi11
closed
3 weeks ago
0
[maca] code refine
#95
Reinerzhou
closed
3 weeks ago
0
[maca] adjust rotary_embedding.
#94
Reinerzhou
closed
3 weeks ago
0
ci:add graph_mode test
#93
JackWeiw
closed
3 weeks ago
0
fix vl prefill_attention
#92
CyCle1024
closed
3 weeks ago
0
[ascend] fix prefill_attention in ascend graph mode
#91
tangzhiyi11
closed
3 weeks ago
1
[ascend] feat: add parallel linear in graph
#90
tangzhiyi11
closed
3 weeks ago
0
ci: only run e2e test on origin repo
#89
CyCle1024
closed
3 weeks ago
0
ci:fix tp model test
#88
JackWeiw
closed
3 weeks ago
0
Bump version to 0.1.1.post2
#87
jinminxi104
closed
4 weeks ago
0
update build wheel scripts
#86
CyCle1024
closed
4 weeks ago
1
update readme
#85
jinminxi104
closed
4 weeks ago
0
ci:fix default eager mode
#84
JackWeiw
closed
4 weeks ago
0
fix: use npu_fusion_attention for prefill
#83
CyCle1024
closed
1 month ago
0
build: build_wheel tools update
#82
CyCle1024
closed
1 month ago
0
bump version to 0.1.1
#81
jinminxi104
closed
1 month ago
0
make compatibility on Ascend310P for MHA models
#80
yao-fengchen
opened
1 month ago
0
feat: make cache_size_limit=256 in dynamo config
#79
CyCle1024
closed
1 month ago
0
[maca] adjust patch paras to support different version of transformers.
#78
Reinerzhou
closed
1 month ago
0
fix ascend_awq
#77
yao-fengchen
closed
1 month ago
0
[ascend] opt: remove print
#76
tangzhiyi11
closed
1 month ago
0
[ascend] feat: support internlm2.5
#75
tangzhiyi11
closed
3 weeks ago
0
Ci/fault test
#74
JackWeiw
closed
3 weeks ago
0
fix: support register_custom_op for more platforms
#73
CyCle1024
closed
1 month ago
0
[ascend] feat: optimize atb codes and support linear op with bias parameter
#72
tangzhiyi11
closed
1 month ago
0
make compatibility for atbgraph mode
#71
yao-fengchen
closed
1 month ago
0
ci:add more models
#70
JackWeiw
closed
4 weeks ago
0
add linear op
#69
yao-fengchen
closed
3 weeks ago
0
[ascend] feat: support internlm2 7b
#68
tangzhiyi11
closed
1 month ago
0
Next