ggerganov/whisper.cpp

Pull Requests Commits

cuda : sync llama.cpp fixes

ggerganov committed 2 years ago

Verified 270b1e48

whisper : enable beam-search by default

ggerganov committed 2 years ago

Verified b7c82a37

whisper : avoid starting sampling threads with bs=1

ggerganov committed 2 years ago

Verified 4c245ea1

whisper : add check for max number of decoders

ggerganov committed 2 years ago

Verified 820f4589

whisper : add comment about the KV cache size

ggerganov committed 2 years ago

Verified 6c8a003a

bench : add batch size 5 bench

ggerganov committed 2 years ago

Verified ae1bd690

whisper : fix decoders initializer

ggerganov committed 2 years ago

Verified 3ed9af34

whisper : speed-up sampling

ggerganov committed 2 years ago

Verified 9006946e

whisper : clear kv cache when using whisper_decode API

ggerganov committed 2 years ago

Verified d7760357

whisper : fix mem leak again + remove oboslete function

ggerganov committed 2 years ago

Verified b2123cb4

whisper : fix memory leak in whisper_batch

ggerganov committed 2 years ago

Verified 3d24e35f

whisper : full batched decoding support

ggerganov committed 2 years ago

Verified 91096daa

whisper : move kv_self to whisper_state

ggerganov committed 2 years ago

Verified 8b943f98

whisper : add whisper_batch

ggerganov committed 2 years ago

Verified 3cbaaed0

java : use tiny.en for tests (#1484)

ggerganov committed 2 years ago

Verified d4231649

whisper : add grammar-based sampling (#1229)

ejones committed 2 years ago

Verified 3e5c7fee

talk-llama : add n_gpu_layers parameter (#1475)

rlapray committed 2 years ago

Verified c23598e4

examples : add whisper.android.java for compatibility with older Android versions using Java (#1382)

litongjava committed 2 years ago

Verified 54a08bde

readme : update comment about source code

ggerganov committed 2 years ago

Verified 9f8bbd3f

ggml : fix some compile warnings

ggerganov committed 2 years ago

Verified 3172006a

readme : update GPU / CUDA

ggerganov committed 2 years ago

Verified 684bc8bd

whisper : add full CUDA and Metal offloading (#1472)

ggerganov committed 2 years ago

Verified b0502836

whisper : return with error from whisper_encode_internal and whisper_decode_internal when abort callback is true (#1456)

bjnortier committed 2 years ago

Verified ec7a6f04

talk-llama : add language auto detect (#1467)

kubaracek committed 2 years ago

Verified 37947203

openvino : update convert-whisper-to-openvino.py to support v3 (#1459)

bobqianic committed 2 years ago

Verified 953419c6

coreml : use the correct `n_mel` value (#1458)

jxy committed 2 years ago

Verified 0de8582f

whisper : reset mel time when resetting timings (#1452)

bjnortier committed 2 years ago

Verified baeb7336

ios : add support for Swift Package Manager (#1370)

sindresorhus committed 2 years ago

Verified d03c60dd

release : v1.4.3

ggerganov committed 2 years ago

Verified 6a5d1951

ggml : fix MIN / MAX macro re-definition

ggerganov committed 2 years ago

Verified 0cbef754

Older