Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
We've been using whisper.cpp (whisper.spm) in www.detail.co and have run into a number of crashes at this assertion:
whisper.cpp/ggml.c
Lines 1653 to 1657 in 0d22916
Debugging these it seems like
exp_n_audio_ctx
is not initialized beforewhisper_lang_auto_detect
->whisper_encode
, andn_ctx
is therefore set to a large random number, creating a mel tensor that runs out of memory.whisper.cpp/whisper.cpp
Line 1368 in 0d22916
Initializing
exp_n_audio_ctx
to 0 fixes this issue for us.