deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-05 18:09:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

mark stateless token2wav

Browse Source
This commit is contained in:
root
2025-09-26 14:51:41 +08:00
parent 482464ea27
commit 31a0adc73d
5 changed files with 266 additions and 121 deletions
Download Patch File Download Diff File Expand all files Collapse all files

80
runtime/triton_trtllm/model_repo/token2wav_dit/config.pbtxt
View File

@@ -15,11 +15,14 @@
name: "token2wav_dit"
backend: "python"
max_batch_size: ${triton_max_batch_size}
dynamic_batching {
max_queue_delay_microseconds: ${max_queue_delay_microseconds}
priority_levels: 10
default_priority_level: 10
}
parameters: { key: "FORCE_CPU_ONLY_INPUT_TENSORS" value: {string_value:"no"}}
parameters [
{
key: "model_dir",
@@ -49,6 +52,48 @@ input [
dims: [ 1 ]
reshape: { shape: [ ] }
optional: true
},
{
name: "conformer_cnn_cache"
data_type: TYPE_FP16
dims: [ 512, -1 ]
optional: true
},
{
name: "conformer_att_cache"
data_type: TYPE_FP16
dims: [ 10, 8, -1, 128 ]
optional: true
},
{
name: "estimator_cnn_cache"
data_type: TYPE_FP16
dims: [ 10, 16, -1, 1024, 2 ]
optional: true
},
{
name: "estimator_att_cache"
data_type: TYPE_FP16
dims: [ 10, 16, -1, 8, -1, 128 ]
optional: true
},
{
name: "mel"
data_type: TYPE_FP32
dims: [ 80, -1 ]
optional: true
},
{
name: "source"
data_type: TYPE_FP32
dims: [ 1, -1 ]
optional: true
},
{
name: "speech"
data_type: TYPE_FP32
dims: [ -1 ]
optional: true
}
]
output [
@@ -56,6 +101,41 @@ output [
name: "waveform"
data_type: TYPE_FP32
dims: [ -1 ]
},
{
name: "conformer_cnn_cache"
data_type: TYPE_FP16
dims: [ 512, -1 ]
},
{
name: "conformer_att_cache"
data_type: TYPE_FP16
dims: [ 10, 8, -1, 128 ]
},
{
name: "estimator_cnn_cache"
data_type: TYPE_FP16
dims: [ 10, 16, -1, 1024, 2 ]
},
{
name: "estimator_att_cache"
data_type: TYPE_FP16
dims: [ 10, 16, -1, 8, -1, 128 ]
},
{
name: "mel"
data_type: TYPE_FP32
dims: [ 80, -1 ]
},
{
name: "source"
data_type: TYPE_FP32
dims: [ 1, -1 ]
},
{
name: "speech"
data_type: TYPE_FP32
dims: [ -1 ]
}
]