OpenNMT/CTranslate2

Issues

• Convert model.bin (fp32) to model.bin (int8)

  #1761 opened 5 months ago
  4

• When I run online for a long time, the gpu memory will get bigger and bigger

  #1759 opened 5 months ago
  0

• set_random_seed does not make temperature based decoding deterministic

  #1757 opened 5 months ago
  4

• Support for ARM64 on Windows

  #1756 opened 5 months ago
  3

• Docker images not published

  #1754 opened 5 months ago
  2

• CI failing in the several recent PRs

  #1753 opened 5 months ago
  2

• Flash Attention regurgitates repeated tokens - seq2seq

  #1752 opened 6 months ago
  1

• Llama 3.1 support please?

  #1745 opened 5 months ago
  2

• Converting to Llama architecture to expand models supported by Ctranslate2

  #1744 opened 4 months ago
  1

• Trouble converting Mistral-Nemo despite architecture indicating MistralForCausalLM

  #1743 opened 4 months ago
  10

• The version of CUB in your include path is not compatible with this release of Thrust.

  #1742 opened 6 months ago
  1

• Introduce better format for whisper models

  #1741 opened 6 months ago
  1

• Other differences in the beam search implementation?

  #1740 opened 6 months ago
  5

• Falcon-11B support

  #1737 opened 7 months ago
  0

• [Feature Request] Expose Profiler to Python

  #1736 opened 7 months ago
  1

• Add gemma2 support

  #1735 opened 3 months ago
  2

• [Profiling] Help needed with Profiling CT2 on both CUDA and CPU

  #1734 opened 7 months ago
  1

• Segmentation fault when using `return_alternatives` flag on NLLB model on cuda

  #1731 opened 7 months ago
  2

• [feature request] Mixed quantizations.

  #1730 opened 7 months ago
  1

• Download ready to use model

  #1729 opened 6 months ago
  3

• Gemma model - help needed

  #1728 opened 7 months ago
  4

• Adding a layer to an existing model?

  #1726 opened 7 months ago
  4

• Feature Request: Support for CohereForAI AYA-23 models

  #1725 opened 7 months ago
  0

• Inference for ctranslate2 using tensor parallel with mpi

  #1724 opened 7 months ago
  1

• Qwen2 Support?

  #1721 opened 7 months ago
  2

• Does CT2/OpenNMT engine support Qualcomm SoC?

  #1720 opened 7 months ago
  1

• T5 inference result is all <pad>

  #1719 opened 7 months ago
  2

• Feature Request: Implement Static Cache and Quantization Techniques in CTranslate2

  #1717 opened 8 months ago
  18

• Proper way to change alignment_heads and alignment_layer when using return_attention=True for converted HF Seq2Seq Transformers?

  #1716 opened 8 months ago
  0

• Facing issues with Ctranslate2 when working with Intel built-in GPU and oneDNN

  #1715 opened 6 months ago
  7

• Ctranslate2 Pypi exceeds limit 20GB

  #1712 opened 5 months ago
  1

• Converter not working for NLLB models

  #1711 opened 8 months ago
  9

• CUDA DeviceAllocate segfault

  #1709 opened 8 months ago
  3

• Different results when run with tensor parallelism

  #1708 opened 8 months ago
  2

• Support for Phi3-Small, Medium, and Vision

  #1707 opened 8 months ago
  2

• Clang unusual switches wrongly hardcoded in resulting setup.py

  #1704 opened 8 months ago
  3

• Doesn't build without docker. libiomp5 not found

  #1703 opened 8 months ago
  6

• Option --self_attn_type scaled-dot-flash is not supported (supported values are: scaled-dot)

  #1702 opened 8 months ago
  6

• Whisper encode roughly 4x slower than openai/pytorch

  #1699 opened 8 months ago
  1

• libctranslate2-81fc0d88.so.4.2.1 in python package has executable stack flag

  #1698 opened 8 months ago
  0

• CTranslate2 cmake error when trying to build the code from source with cuda support enabled on Windows.

  #1697 opened 8 months ago
  6

• I got invalid conversion error when compile on linux

  #1696 opened 8 months ago
  2

• opus-mt-en-zh does not respect the end token

  #1694 opened 8 months ago
  0

• Can't hide GPUs to get_cuda_device_count()

  #1693 opened 8 months ago
  5

• How to compile from source on windows 11?

  #1692 opened 9 months ago
  3

• Unexpected inference results from Flan-T5 XXL converted to ctranslate2 with version 4.2.1 and 4.1.1 (using tensor parallel)

  #1691 opened 9 months ago
  4

• target_prefix latency

  #1689 opened 9 months ago
  2

• [SOLVED] Running Llama3 with Ctranslate2

  #1688 opened 3 months ago
  4

• Dynamic LoRA switching

  #1686 opened 9 months ago
  2

• Problem converting Phi3-instruct-128k; "su" rope scaling in Phi-3

  #1685 opened 8 months ago
  3