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

fix lint

Browse Source
This commit is contained in:
lyuxiang.lx
2025-02-06 16:07:13 +08:00
parent 24f796a2b1
commit 2a3e033ee1
17 changed files with 187 additions and 135 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
cosyvoice/bin/export_trt.sh
View File

@@ -7,19 +7,19 @@ MODEL_DIR=<YOUR_MODEL_DIR>
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$TRT_DIR/lib:/usr/local/cuda/lib64
# cosyvoice export
$TRT_DIR/bin/trtexec --onnx=$MODEL_DIR/flow.decoder.estimator.fp32.onnx --saveEngine=$MODEL_DIR/flow.decoder.estimator.fp32.mygpu.plan --minShapes=x:2x80x4,mask:2x1x4,mu:2x80x4,cond:2x80x4 --optShapes=x:2x80x193,mask:2x1x193,mu:2x80x193,cond:2x80x193 --maxShapes=x:2x80x6800,mask:2x1x6800,mu:2x80x6800,cond:2x80x6800 --inputIOFormats=fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw --outputIOFormats=fp32:chw
$TRT_DIR/bin/trtexec --onnx=$MODEL_DIR/flow.decoder.estimator.fp32.onnx --saveEngine=$MODEL_DIR/flow.decoder.estimator.fp16.mygpu.plan --fp16 --minShapes=x:2x80x4,mask:2x1x4,mu:2x80x4,cond:2x80x4 --optShapes=x:2x80x193,mask:2x1x193,mu:2x80x193,cond:2x80x193 --maxShapes=x:2x80x6800,mask:2x1x6800,mu:2x80x6800,cond:2x80x6800 --inputIOFormats=fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw --outputIOFormats=fp16:chw
$TRT_DIR/bin/trtexec --onnx=$MODEL_DIR/flow.decoder.estimator.fp32.onnx --saveEngine=$MODEL_DIR/flow.decoder.estimator.fp32.mygpu.plan --minShapes=x:2x80x4,mask:2x1x4,mu:2x80x4,cond:2x80x4 --optShapes=x:2x80x200,mask:2x1x200,mu:2x80x200,cond:2x80x200 --maxShapes=x:2x80x3000,mask:2x1x3000,mu:2x80x3000,cond:2x80x3000 --inputIOFormats=fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw --outputIOFormats=fp32:chw
$TRT_DIR/bin/trtexec --onnx=$MODEL_DIR/flow.decoder.estimator.fp32.onnx --saveEngine=$MODEL_DIR/flow.decoder.estimator.fp16.mygpu.plan --fp16 --minShapes=x:2x80x4,mask:2x1x4,mu:2x80x4,cond:2x80x4 --optShapes=x:2x80x200,mask:2x1x200,mu:2x80x200,cond:2x80x200 --maxShapes=x:2x80x3000,mask:2x1x3000,mu:2x80x3000,cond:2x80x3000 --inputIOFormats=fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw --outputIOFormats=fp16:chw
# cosyvoice2 export with cache
$TRT_DIR/bin/trtexec --onnx=$MODEL_DIR/flow.decoder.estimator.fp32.onnx --saveEngine=$MODEL_DIR/flow.decoder.estimator.fp32.mygpu.plan \
--minShapes=x:2x80x4,mask:2x1x4,mu:2x80x4,cond:2x80x4,down_blocks_kv_cache:1x4x2x0x512x2,mid_blocks_kv_cache:12x4x2x0x512x2,up_blocks_kv_cache:1x4x2x0x512x2 \
--optShapes=x:2x80x193,mask:2x1x193,mu:2x80x193,cond:2x80x193,down_blocks_kv_cache:1x4x2x193x512x2,mid_blocks_kv_cache:12x4x2x193x512x2,up_blocks_kv_cache:1x4x2x193x512x2 \
--maxShapes=x:2x80x6800,mask:2x1x6800,mu:2x80x6800,cond:2x80x6800,down_blocks_kv_cache:1x4x2x200x512x2,mid_blocks_kv_cache:12x4x2x200x512x2,up_blocks_kv_cache:1x4x2x200x512x2 \
--optShapes=x:2x80x200,mask:2x1x200,mu:2x80x200,cond:2x80x200,down_blocks_kv_cache:1x4x2x100x512x2,mid_blocks_kv_cache:12x4x2x100x512x2,up_blocks_kv_cache:1x4x2x100x512x2 \
--maxShapes=x:2x80x1500,mask:2x1x1500,mu:2x80x1500,cond:2x80x1500,down_blocks_kv_cache:1x4x2x200x512x2,mid_blocks_kv_cache:12x4x2x200x512x2,up_blocks_kv_cache:1x4x2x200x512x2 \
--inputIOFormats=fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw \
--outputIOFormats=fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw,fp32:chw
$TRT_DIR/bin/trtexec --onnx=$MODEL_DIR/flow.decoder.estimator.fp32.onnx --saveEngine=$MODEL_DIR/flow.decoder.estimator.fp16.mygpu.plan --fp16 \
--minShapes=x:2x80x4,mask:2x1x4,mu:2x80x4,cond:2x80x4,down_blocks_kv_cache:1x4x2x0x512x2,mid_blocks_kv_cache:12x4x2x0x512x2,up_blocks_kv_cache:1x4x2x0x512x2 \
--optShapes=x:2x80x193,mask:2x1x193,mu:2x80x193,cond:2x80x193,down_blocks_kv_cache:1x4x2x193x512x2,mid_blocks_kv_cache:12x4x2x193x512x2,up_blocks_kv_cache:1x4x2x193x512x2 \
--maxShapes=x:2x80x6800,mask:2x1x6800,mu:2x80x6800,cond:2x80x6800,down_blocks_kv_cache:1x4x2x200x512x2,mid_blocks_kv_cache:12x4x2x200x512x2,up_blocks_kv_cache:1x4x2x200x512x2 \
--optShapes=x:2x80x200,mask:2x1x200,mu:2x80x200,cond:2x80x200,down_blocks_kv_cache:1x4x2x100x512x2,mid_blocks_kv_cache:12x4x2x100x512x2,up_blocks_kv_cache:1x4x2x100x512x2 \
--maxShapes=x:2x80x1500,mask:2x1x1500,mu:2x80x1500,cond:2x80x1500,down_blocks_kv_cache:1x4x2x200x512x2,mid_blocks_kv_cache:12x4x2x200x512x2,up_blocks_kv_cache:1x4x2x200x512x2 \
--inputIOFormats=fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw \
--outputIOFormats=fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw,fp16:chw