Repository Issue Activity (beta)

#2101 ROCm/HIP Windows wheel: Translator destruction hangs indefinitely, even in CPU-only mode with no AMD GPU present

#1933 Feature request: CUDA 13 support

#2100 Significant Korean accuracy degradation with Whisper inputs shorter than 30 seconds

#2088 Host memory grows ~6.8 KB per inference on aarch64 + CUDA (not on x86_64, not on CPU)

#2093 Chutki

#2090 HIP: hipMallocAsync silently corrupts buffers on gfx1030 (RDNA2) although the device reports memory-pool support; default allocator should be cub_caching on HIP

#2089 Removed by author

#1950 Making it easier for users to get started with CTranslate2 models

#2079 import ctranslate2 eagerly imports torch (via converters + specs), even for inference-only use

#1782 Python process crashes on exit under Windows with CUDA

#2085 Windows 11 process-exit hang after ROCm Whisper inference with AMD GPU

#2078 import ctranslate2 eagerly imports torch (via converters + specs), even for inference-only use

#2082 Source builds need workarounds for sdist, CUDA arches, non-x86 CPU dispatch, and setup.py hardcoding

#2021 ROCm (gfx1201/RX 9070 XT): faster-whisper crash in CTranslate2 4.7.1

#2081 Failure to compile on 32-bit x86 GNU/Linux

#1181 Importing `numpy` after `ctranslate2` causes segfault

#2063 macOS Python 3.14 CPU backend: intra_threads=0 aborts on 4.7.2 and oversubscribes on 4.8.0

#1607 MPS Support for Mac Apple Silicon?

#2074 Allow no-timestamp Whisper decoding beyond 224 tokens

#2072 Error on using converted model

#2034 Private security contact request

#1765 Reintroduce support for Compute Capability 5.0

#2064 4.8.0 regression: WhisperModel.align() fails with parallel_for failed: cudaErrorInvalidDevice on Tesla T4 (works on 4.7.0)

#518 Alpine support

#1946 T5Gemma