 AlpinDale
|
5c3b94de45
spec decode: move ops.advane_step to flash attention backend (#1005)
|
2 viikkoa sitten |
|
AlpinDale
|
3bb0f07461
chore: rename `task_handler` to `worker` (#985)
|
2 viikkoa sitten |
|
AlpinDale
|
0dfa6b60ec
core: support logprobs with multi-step scheduling (#963)
|
2 viikkoa sitten |
|
AlpinDale
|
132aa2abe4
spec decode: add support for EAGLE (#899)
|
3 viikkoa sitten |
|
AlpinDale
|
1405051912
attention: add `AttentionState` abstraction (#863)
|
1 kuukausi sitten |
|
AlpinDale
|
2f61644f6e
SPMD optimizations (#824)
|
1 kuukausi sitten |
|
AlpinDale
|
89a2c6dee1
chore: refactor `MultiModalConfig` initialization and profiling (#745)
|
3 kuukautta sitten |
|
AlpinDale
|
7a313483f1
chore: move update_flash_attn_metadata to attn backend (#731)
|
3 kuukautta sitten |
|
AlpinDale
|
f1d0b77c92
[0.6.0] Release Candidate (#481)
|
4 kuukautta sitten |
