Georgi Gerganov
b41f4a90eb
Create README.md
2 years ago
Georgi Gerganov
bb1ee266d2
ios : whisper.objc example
2 years ago
Georgi Gerganov
5f7e9fa2dc
ref #68 , #79 : fix segment time output
2 years ago
Georgi Gerganov
181b762de8
Update README.md
2 years ago
Georgi Gerganov
3d37ad5133
Merge pull request #78 from jokkebk/Specify-utf8-for-vocab.json
...
Add enconding parameter to vocab.json opening to fix errors
2 years ago
Joonas Pihlajamaa
4e887dc350
Add enconding parameter to vocab.json opening to fix errors
2 years ago
Georgi Gerganov
4196856c7b
Update README.md
2 years ago
Georgi Gerganov
705198f063
Update README.md
2 years ago
Georgi Gerganov
3e69a6071d
Update README.md
2 years ago
Georgi Gerganov
f3dae90c31
Update README.md
2 years ago
Georgi Gerganov
6a81ed3e78
main : print colors + no timestamps
2 years ago
Georgi Gerganov
7affd309d3
whisper : add new-segment callback
...
Can be used to process new segments as they are being generated.
Sample usage in main, for printing the resulting segments during the
inference.
2 years ago
Georgi Gerganov
8f95c25aed
main : refactor subtitle output
2 years ago
Georgi Gerganov
31ff0c6a1f
wip : experimental color coding of tokens based on probabilities
2 years ago
Georgi Gerganov
f4aa01c2f8
Update README.md
2 years ago
Georgi Gerganov
8c1d970088
Update README.md
2 years ago
Georgi Gerganov
6b45e37b2b
Update README.md and finalize the whisper.wasm example
2 years ago
Georgi Gerganov
491ecd7056
wip : polishing WASM example
2 years ago
Georgi Gerganov
db460b78ff
wip : WASM 128-bit SIMD support
2 years ago
Georgi Gerganov
e905c6f827
wip : initial WASM port
...
Works but it is very slow because no SIMD is used.
For example, jfk.wav is processed in ~23 seconds using "tiny.en" model
2 years ago
Georgi Gerganov
7d0dee7a8a
ref #68 : add option "-on" to specify segment index offset for SRT
...
Also, change option "-o" to "-ot"
2 years ago
Georgi Gerganov
8d15a1c635
ci : fix and re-enable tests (2nd try)
2 years ago
Georgi Gerganov
692aa0784f
Revert "ci : fix and re-enable tests"
...
This reverts commit 80aefc9514
.
2 years ago
Georgi Gerganov
80aefc9514
ci : fix and re-enable tests
2 years ago
Georgi Gerganov
5698b51718
Update README.md
2 years ago
Georgi Gerganov
3fe3898ebb
Update README.md
2 years ago
Georgi Gerganov
81c185576c
Update README.md
2 years ago
Georgi Gerganov
744bd47685
Merge pull request #67 from undefdev/defensive-apple-arm-make
...
added handling for falsely as x86_64 announced ARM Macs
2 years ago
Georgi Gerganov
66b3169d39
ci : disable tests temporarily
2 years ago
undef
19a780afe5
added handling for falsely as x86_64 announced ARM Macs
2 years ago
Georgi Gerganov
1969ee4bc7
Update README.md
2 years ago
Georgi Gerganov
0e4fd43400
stream : print warning when processing is not fast enough
2 years ago
Georgi Gerganov
19817711b4
Add reference to FP16 repo
2 years ago
Georgi Gerganov
7eeef0358a
ref #52 : improve greedy sampling strategy
...
Force timestamp token to be sampled if the probability sum over all
timestamp tokens is above the probability of any other token
2 years ago
Georgi Gerganov
632660abb9
CMake support for Accelerate framework
2 years ago
Georgi Gerganov
e36aabe00d
Correct implementation of FP16 GELU
...
Can toggle it via the GGML_GELU_FP16 macro
2 years ago
Georgi Gerganov
2d171ced32
close #32 : add comment about thread-safety of the C-style API
2 years ago
Georgi Gerganov
e30cf83158
ref #57 , #62 , #63 : remove unions in C-api + remove designated initializers
...
We are not ready for designated initializers - many compilers do not
support this C++ feature yet, so removing it's non-trivial usages.
2 years ago
Georgi Gerganov
d6b84b2a23
ref #62 : fix build for some compilers
...
For some reason, new version of GCC panic when the struct type is not
specified explicitly
2 years ago
Georgi Gerganov
b4a3875b2c
Revert recent sampling change
...
It does not actually help and seems to produce worse results on some of
the samples
2 years ago
Georgi Gerganov
cf67bfffa0
Fix EOT token handling
...
If it is the end of the audio, pick all sampled tokens.
Otherwise, print error message.
2 years ago
Georgi Gerganov
91632eb6ea
Revert GELU change
...
Seems it does not work on x86 for some reason
2 years ago
Georgi Gerganov
b81a81d543
Link Accelerate framework to "stream" example
2 years ago
Georgi Gerganov
d14823582d
Try to improve the sampling strategy a bit
...
It sill fails sometimes when it does not sample a timestamp token for
the entire segment. We now print a message in such cases
2 years ago
Georgi Gerganov
20d8e7a309
Fix memory sizes
2 years ago
Georgi Gerganov
72d967bce4
Use Accelerate framework on Apple silicon
...
Huge performance improvement in the Encode (almost x2 on MacBook M1 Pro)
Also various extra optimizations:
- Multi-threaded NORM operator
- Faster GELU via F16 cast
2 years ago
Georgi Gerganov
130b5c02d6
Adding helper script for converting the PT models
2 years ago
Georgi Gerganov
0e858f080d
close #56 : build on FreeBSD
...
Thanks to @abelbabel for the contribution
2 years ago
Georgi Gerganov
f24d940ca9
Merge pull request #58 from r0y6a3n0/master
...
fix decode missing token issue
2 years ago
RyanChang
949f97a8b4
fix missing token issue
2 years ago