 AlpinDale
|
e177788401
add moe tests
|
vor 9 Monaten |
|
AlpinDale
|
9da2946652
update kernel tests
|
vor 9 Monaten |
|
AlpinDale
|
2172a9c374
add fp8_e4m3fn scales for llama2 7b and 70b
|
vor 9 Monaten |
|
AlpinDale
|
e28c8496b2
endpoint tests
|
vor 9 Monaten |
|
AlpinDale
|
ce10891496
add more engine-related tests:
|
vor 9 Monaten |
|
AlpinDale
|
bfca95c1a8
update detokenization test
|
vor 9 Monaten |
|
AlpinDale
|
9e1cea354c
add distributed system tests
|
vor 9 Monaten |
|
AlpinDale
|
54ed7adef3
add core processor tests
|
vor 9 Monaten |
|
AlpinDale
|
3a206f9e11
add chunked prefill correctness test
|
vor 9 Monaten |
|
AlpinDale
|
0558b22749
basic correctness test
|
vor 9 Monaten |
|
AlpinDale
|
9082ac7b7a
add async engine test units
|
vor 9 Monaten |
 sgsdxzy
|
fcfb72af24
Support arbitrary model in GGUF. (#381)
|
vor 9 Monaten |
 AlpinDale
|
bd0ddf1cfe
feat: EETQ quantization (#408)
|
vor 9 Monaten |
|
AlpinDale
|
b1caee23a6
cache the p2p access check for memory saving
|
vor 9 Monaten |
|
AlpinDale
|
373e0d3c01
fix neuron
|
vor 9 Monaten |
|
AlpinDale
|
28bcca2396
incorrect use of monotonic time in metrics logger
|
vor 9 Monaten |
|
AlpinDale
|
4ba273886a
debug logging for distributed_init_method
|
vor 9 Monaten |
|
AlpinDale
|
1270b5567e
triton compile error for flash_attn
|
vor 9 Monaten |
|
AlpinDale
|
f375353026
enable custom_all_reduce by default in llm.py
|
vor 9 Monaten |
|
AlpinDale
|
2d2b43fe00
fix type hint
|
vor 9 Monaten |
|
AlpinDale
|
531969a0b2
move merge_async_iterators to common utils
|
vor 9 Monaten |
|
AlpinDale
|
c18bf116da
fix stop strings not being excluded from outputs
|
vor 9 Monaten |
|
AlpinDale
|
5ab7a159d7
fix formatting for previous commit
|
vor 9 Monaten |
|
AlpinDale
|
b6bbf584ac
fix echo
|
vor 9 Monaten |
|
AlpinDale
|
6e0761ba5d
make init_distributed_environment compatible with init_process_group
|
vor 9 Monaten |
|
AlpinDale
|
083ba7b452
roll back chunked prefill changes to SDPA, isolate cpu worker
|
vor 9 Monaten |
|
AlpinDale
|
8c67b37131
fix docstrings
|
vor 9 Monaten |
|
AlpinDale
|
fe17712f29
fully working chunked prefill
|
vor 9 Monaten |
|
AlpinDale
|
8db2fa8e2e
why was that not committed?
|
vor 9 Monaten |
|
AlpinDale
|
54678c91f3
fix outlines requirements
|
vor 9 Monaten |
