.. |
adapter_commons
|
2f61644f6e
SPMD optimizations (#824)
|
преди 3 месеца |
assets
|
653d1a08d4
feat: add support for audio models (#891)
|
преди 2 месеца |
attention
|
d6cbbba95f
Revert "fix: issues with flashinfer fp8 kv (#950)" (#956)
|
преди 2 месеца |
common
|
5bec8fbb1b
tpu: add support for async postprocessing (#968)
|
преди 2 месеца |
compilation
|
0e5cf7f840
tpu: avoid dynamo guard eval overhead (#949)
|
преди 2 месеца |
distributed
|
2940da2c7b
distributed: fix custom allreduce p2p cache file generation (#922)
|
преди 2 месеца |
endpoints
|
39b2e83ac3
api: optimize zeromq frontend performance (#951)
|
преди 2 месеца |
engine
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
преди 2 месеца |
executor
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
преди 2 месеца |
inputs
|
908ff753a1
fix: phi_3.5_v loading (#896)
|
преди 2 месеца |
kv_quant
|
8a71788372
Add OLMoE (#772)
|
преди 5 месеца |
lora
|
bf4a4d8516
fix: do not register punica with torch if using older torch (#948)
|
преди 2 месеца |
modeling
|
a8bdd488b9
distributed: support pipeline parallelism for internvl and internlm2 (#965)
|
преди 2 месеца |
multimodal
|
2aabf8fcf7
vlm: fix errors on ragged NestedTensors (#953)
|
преди 2 месеца |
platforms
|
9f3e7c86e2
feat: add fused Marlin MoE kernel (#934)
|
преди 2 месеца |
plugins
|
22a4cd4595
core: fix spec decode metrics and envs circular import (#889)
|
преди 2 месеца |
processing
|
5be6225f38
core: support multi-step scheduling w/ async post-processor (#955)
|
преди 2 месеца |
prompt_adapter
|
2f61644f6e
SPMD optimizations (#824)
|
преди 3 месеца |
quantization
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
преди 2 месеца |
server
|
22a4cd4595
core: fix spec decode metrics and envs circular import (#889)
|
преди 2 месеца |
spec_decode
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
преди 2 месеца |
task_handler
|
5bec8fbb1b
tpu: add support for async postprocessing (#968)
|
преди 2 месеца |
transformers_utils
|
9f3e7c86e2
feat: add fused Marlin MoE kernel (#934)
|
преди 2 месеца |
triton_utils
|
f1d0b77c92
[0.6.0] Release Candidate (#481)
|
преди 6 месеца |
__init__.py
|
f1d0b77c92
[0.6.0] Release Candidate (#481)
|
преди 6 месеца |
_core_ext.py
|
f1ea7711bd
core: do not compile ScalarType for torch < 2.4.0 (#938)
|
преди 2 месеца |
_custom_ops.py
|
fcfcfc65e1
quants: add triton kernels for AWQ (#946)
|
преди 2 месеца |
_ipex_ops.py
|
f1d0b77c92
[0.6.0] Release Candidate (#481)
|
преди 6 месеца |
connections.py
|
c6c91edab7
ci: update & overhaul test units (#769)
|
преди 3 месеца |
constants.py
|
2f61644f6e
SPMD optimizations (#824)
|
преди 3 месеца |
py.typed
|
1c988a48b2
fix logging and add py.typed
|
преди 1 година |
scalar_type.py
|
f1d0b77c92
[0.6.0] Release Candidate (#481)
|
преди 6 месеца |
version.py
|
cbd51a208a
ci: bump to 0.6.5 (#964)
|
преди 2 месеца |