Reduce memory usage during Whisper inference #431
ggerganov
force pushed
from
dadead51
to
34b8afe6
2 years ago
ggerganov
force pushed
from
50692634
to
f1b9913b
2 years ago
ggerganov
force pushed
from
1f7cd041
to
60d0f9da
2 years ago
ggerganov
marked this pull request as ready for review 2 years ago
ggerganov
force pushed
from
b0c2268f
to
41f31719
2 years ago
ggml : add "scratch" buffer support
60eff46b
ggml : support for scratch ring-buffer
0eea547a
ggml : bug fix in ggml_repeat()
18210579
ggml : error on scratch buffer overflow
0ba91b54
whisper : use scratch buffers during inference (base model only)
1a1dee46
whisper : update memory usage for all models
6cae05bd
whisper : fix encoder memory usage
79148a21
whisper : use whisper_context functions instead of macros
42d7dee4
whisper : fix FF + remove it from README
4e0e2520
ggml : reuse ggml_new_i32
d922aa4a
ggml : refactor the scratch buffer storage
62205aed
whisper : reorder scratch buffers in the decoder
01669ee8
main : add option to disable temp fallback
bdf21fa6
Update README.md
6ed13449
ggerganov
force pushed
from
41f31719
to
6ed13449
2 years ago
js : update whisper.js
a26f3a71
ggerganov
merged
f3ee4a96
into master 2 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub