50h100a
|
f67b5be198
chore: port sampler+metadata changes from main to dev (#427)
|
9 meses atrás |
AlpinDale
|
2319b411ce
refactor: neuron support
|
9 meses atrás |
AlpinDale
|
9181fa0396
feat: Triton kernels for sampling (#383)
|
9 meses atrás |
AlpinDale
|
f8dfac6372
chore: attention refactor and upstream sync apr01 (#365)
|
9 meses atrás |
AlpinDale
|
9fa99215f8
feat: add cubic sampling (#280)
|
10 meses atrás |
AlpinDale
|
657aec0cbd
refactor: OpenAI endpoint (#261)
|
10 meses atrás |
AlpinDale
|
4d04ade9ef
feat: fine-grained seeds (#279)
|
11 meses atrás |
AlpinDale
|
1c46fa31ad
feat: add quadratic sampling (#233)
|
11 meses atrás |
AlpinDale
|
8fa608aeb7
feat: replace Ray with NCCL for control plane comms (#221)
|
1 ano atrás |
AlpinDale
|
a39eeb7188
fix: logprobs for dynatemp (#215)
|
1 ano atrás |
Stefan Gligorijevic
|
56446a04bb
feat: dynamic temperature (#209)
|
1 ano atrás |
AlpinDale
|
1394eab8ab
fix temperature being set to 1 in all cases (#210)
|
1 ano atrás |
AlpinDale
|
d54791aaa8
feat: reduce sampler overhead by making it less blocking (#198)
|
1 ano atrás |
g4rg
|
2aab3da9bd
chore: fix Python 3.8+ compatibility (#170)
|
1 ano atrás |
AlpinDale
|
653da510d1
chore: rewrite InputMetadata (#143)
|
1 ano atrás |