vllm-project/vllm

Pull Requests Commits
Add buffer donation to benchmark
WoosukKwon committed 1 year ago
c00ddd68
Add block size
WoosukKwon committed 1 year ago
881b8840
Disable memory tracking
WoosukKwon committed 1 year ago
98a3df0f
Fix for binary cache
WoosukKwon committed 1 year ago
3f6288cc
Tune pages_per_compute_block
WoosukKwon committed 1 year ago
408ff495
Add tpu
WoosukKwon committed 1 year ago
278e8a1a
Improve benchmark
WoosukKwon committed 1 year ago
07be6ed3
Use persistent cache
WoosukKwon committed 1 year ago
f6637dba
Move JAX-smi to worker
WoosukKwon committed 1 year ago
707a5f64
Fix bucketing
WoosukKwon committed 1 year ago
57690a9c
Add precompilation step
WoosukKwon committed 1 year ago
b15db234
Add op benchmark scripts
WoosukKwon committed 1 year ago
d1591f0f
yapf
WoosukKwon committed 1 year ago
85d44884
yapf
WoosukKwon committed 1 year ago
8d072dbf
yapf
WoosukKwon committed 1 year ago
d830766c
Add warmup + formatting
WoosukKwon committed 1 year ago
5ae2f81c
yapf
WoosukKwon committed 1 year ago
4ea41d01
Add comment
WoosukKwon committed 1 year ago
d16a3484
Format gemma.py
WoosukKwon committed 1 year ago
aa092834
Fix is_tpu
WoosukKwon committed 1 year ago
d2c6a32c
Change version
WoosukKwon committed 1 year ago
21f35c22
Minor
WoosukKwon committed 1 year ago
2aa9831d
Fix KV cache shape
WoosukKwon committed 1 year ago
028f528a
Add warmup
WoosukKwon committed 1 year ago
fa5bacd5
Fix scheduler
WoosukKwon committed 1 year ago
b62170e4
Add timer
WoosukKwon committed 1 year ago
98eda578
Pad to avoid recompilation
WoosukKwon committed 1 year ago
81b8b813
Minor
WoosukKwon committed 1 year ago
e2c7dedb
Increase #blocks
WoosukKwon committed 1 year ago
5323969f
Include argmax to jit
WoosukKwon committed 1 year ago
f42b4c27
Terms Of Use Privacy Policy Refund Policy Impressum