Offline Qwen3-ASR model configuration.
More...
#include <cxx-api.h>
Definition at line 598 of file cxx-api.h.
◆ conv_frontend
| std::string sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::conv_frontend |
Conv-frontend ONNX model file.
Definition at line 600 of file cxx-api.h.
◆ decoder
| std::string sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::decoder |
Decoder ONNX model file (KV cache).
Definition at line 604 of file cxx-api.h.
◆ encoder
| std::string sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::encoder |
Encoder ONNX model file.
Definition at line 602 of file cxx-api.h.
◆ hotwords
| std::string sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::hotwords |
Optional comma-separated hotwords (UTF-8, ASCII ','), e.g. "foo,bar,baz".
Definition at line 609 of file cxx-api.h.
◆ max_new_tokens
| int32_t sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::max_new_tokens = 128 |
Maximum number of new tokens to generate.
Definition at line 613 of file cxx-api.h.
◆ max_total_len
| int32_t sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::max_total_len = 512 |
Maximum total sequence length supported by the model.
Definition at line 611 of file cxx-api.h.
◆ seed
| int32_t sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::seed = 42 |
Random seed for reproducible sampling.
Definition at line 619 of file cxx-api.h.
◆ temperature
| float sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::temperature = 1e-6f |
Sampling temperature.
Definition at line 615 of file cxx-api.h.
◆ tokenizer
| std::string sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::tokenizer |
Tokenizer directory (e.g. containing vocab.json).
Definition at line 606 of file cxx-api.h.
◆ top_p
| float sherpa_onnx::cxx::OfflineQwen3ASRModelConfig::top_p = 0.8f |
Top-p (nucleus) sampling parameter.
Definition at line 617 of file cxx-api.h.
The documentation for this struct was generated from the following file:
Collaboration diagram for sherpa_onnx::cxx::OfflineQwen3ASRModelConfig:
