 AlpinDale
|
6a57861fca
feat: initial XPU support via intel_extension_for_pytorch (#571)
|
7 bulan lalu |
 AlpinDale
|
a89c9a0e92
fix: device ordinal issues with world_size and stuff
|
7 bulan lalu |
|
AlpinDale
|
fe21123a1c
feat: TPU support (#570)
|
7 bulan lalu |
|
AlpinDale
|
156f577f79
feat: switch from `PYBIND11_MODULE` to `TORCH_LIBRARY` (#569)
|
7 bulan lalu |
|
AlpinDale
|
b029a544ff
optimize eager mode host time with numpy
|
7 bulan lalu |
|
AlpinDale
|
f2b7a42c4e
fix: async cancels in merge_async_iterators for python>=3.9
|
7 bulan lalu |
|
AlpinDale
|
7194047318
remove vllm-nccl
|
7 bulan lalu |
|
AlpinDale
|
90ceab32ff
refactor: consolidate prompt args to LLM engines
|
7 bulan lalu |
|
AlpinDale
|
656459fd84
make fp8_e4m3 work on nvidia
|
8 bulan lalu |
|
AlpinDale
|
251568470e
initial nvidia fp8 e4m3 for kv cache
|
8 bulan lalu |
|
AlpinDale
|
4476d2d885
remove cuda version check
|
8 bulan lalu |
|
AlpinDale
|
2351a0e2cd
feat: FlashInfer backend for decoding phase (#548)
|
8 bulan lalu |
|
AlpinDale
|
2656df543b
why was this removed? weird
|
8 bulan lalu |
|
AlpinDale
|
2e0b115ce1
move func tracing to utils
|
8 bulan lalu |
|
AlpinDale
|
46159b107a
formatting: pt1
|
8 bulan lalu |
|
AlpinDale
|
fca911ee0a
vLLM Upstream Sync (#526)
|
8 bulan lalu |
|
AlpinDale
|
f894f7b176
Revert "reduce dedupe by wrapping in general worker class"
|
10 bulan lalu |
|
AlpinDale
|
9fff6fb507
reduce dedupe by wrapping in general worker class
|
10 bulan lalu |
|
AlpinDale
|
9d81716bfd
[v0.5.3] Release Candidate (#388)
|
10 bulan lalu |
|
AlpinDale
|
e3252edd07
fix: remove event and stream, add typing (#382)
|
11 bulan lalu |
|
AlpinDale
|
33b3786175
fix: cache neuron checks (#379)
|
11 bulan lalu |
|
AlpinDale
|
f8dfac6372
chore: attention refactor and upstream sync apr01 (#365)
|
11 bulan lalu |
|
AlpinDale
|
e53842bd5d
fix: cuda home detection for fp8 kv cache
|
1 tahun lalu |
|
AlpinDale
|
e42a78381a
feat: switch from pylint to ruff (#322)
|
1 tahun lalu |
|
AlpinDale
|
c2d77b1822
chore: logging refactor (#302)
|
1 tahun lalu |
|
AlpinDale
|
9810daa699
feat: INT8 KV Cache (#298)
|
1 tahun lalu |
|
AlpinDale
|
ea0f57b233
feat: allow further support for non-cuda devices (#247)
|
1 tahun lalu |
|
AlpinDale
|
31c95011a6
feat: FP8 E5M2 KV Cache (#226)
|
1 tahun lalu |
|
AlpinDale
|
c0aac15421
feat: S-LoRA support (#222)
|
1 tahun lalu |
|
AlpinDale
|
8fa608aeb7
feat: replace Ray with NCCL for control plane comms (#221)
|
1 tahun lalu |
