Uncaught exception (regex)

[2haloes]

Hello,

I'm assuming that I'm not using this on a fully supported system but I cannot get this to run on Android with Termux.

I generated a q4_1 version of the standard 1.5 model with the Python script and followed the basic guide to setting everything up.

~/stable-diffusion.cpp/build $ ./bin/sd -m ../models/v1-5-pruned-emaonly-ggml-model-q4_1.bin -p "A cityscape at sunset, oil painting" -v                                
Option:                                                     
n_threads:       4                                      
mode:            txt2img                                
model_path:      ../models/v1-5-pruned-emaonly-ggml-model-q4_1.bin                                              
output_path:     output.png                             
init_img:                                               
prompt:          A cityscape at sunset, oil painting   
negative_prompt:                                        
cfg_scale:       7.00                                   
width:           512                                    
height:          512                                    
sample_method:   eular a                                
sample_steps:    20                                     
strength:        0.75                                   
seed:            42                                 
System Info:                                                
BLAS = 0                                                
SSE3 = 0                                                
AVX = 0                                                 
AVX2 = 0                                                
AVX512 = 0                                              
AVX512_VBMI = 0                                         
AVX512_VNNI = 0                                         
FMA = 0                                                 
NEON = 1                                                
ARM_FMA = 1                                             
F16C = 0                                                
FP16_VA = 0                                             
WASM_SIMD = 0                                           
VSX = 0                                             
[INFO]  stable-diffusion.cpp:2687 - loading model from '../models/v1-5-pruned-emaonly-ggml-model-q4_1.bin'      
[DEBUG] stable-diffusion.cpp:2695 - verifying magic     
[DEBUG] stable-diffusion.cpp:2706 - loading hparams     
[INFO]  stable-diffusion.cpp:2712 - ftype: q4_1         
[DEBUG] stable-diffusion.cpp:2718 - loading vocab       
[DEBUG] stable-diffusion.cpp:2746 - ggml tensor size = 272 bytes                                                
[DEBUG] stable-diffusion.cpp:2751 - clip params ctx size =  75.02 MB                                            
[DEBUG] stable-diffusion.cpp:2770 - unet params ctx size =  1287.24 MB                                          
[DEBUG] stable-diffusion.cpp:2791 - vae params ctx size =  95.51 MB                                             
[DEBUG] stable-diffusion.cpp:2812 - preparing memory for the weights                                            
[DEBUG] stable-diffusion.cpp:2828 - loading weights     
[DEBUG] stable-diffusion.cpp:2932 - model size = 1454.34MB                                                      
[INFO]  stable-diffusion.cpp:2941 - total params size = 1454.64MB (clip 73.80MB, unet 1286.34MB, vae 94.51MB)   
[INFO]  stable-diffusion.cpp:2943 - loading model from '../models/v1-5-pruned-emaonly-ggml-model-q4_1.bin' completed, taking 1.32s                                      
terminating with uncaught exception of type std::__ndk1::regex_error: The parser did not consume the entire regular expression.                                         
Aborted

[Green-Sky]

Looks like the android stl has errors, or we use the regex functionality wrong.

[Green-Sky]

@2haloes can your try changing the value here
https://github.com/leejet/stable-diffusion.cpp/blob/master/CMakeLists.txt#L38C1-L38C53
to cxx_std_17 ?

[2haloes]

I changed the value and built from scratch but I get the same exception

[ClashSAN]

#16 solves the problem

[2haloes]

I've tried with the PR and the issue still occurs with Termux

[leejet]

I have merged #16 and made some modifications. You can try pulling the latest master branch to see the changes.

[ClashSAN]

with the modifications, its still working like normal for me, try "q4_0"?

Details

~/stable-diffusion.cpp/build5 $ cp bin/sd ../sd3~/stable-diffusion.cpp/build5 $ cd ..
~/stable-diffusion.cpp $ ./sd3 -s 1 -m anything-v3-1-ggml-model-q4_0.bin -H 256 -W 192 -p "cat_ears, kimono" --steps 10
[INFO]  stable-diffusion.cpp:2687 - loading model from 'anything-v3-1-ggml-model-q4_0.bin'
[INFO]  stable-diffusion.cpp:2712 - ftype: q4_0
[INFO]  stable-diffusion.cpp:2941 - total params size = 1431.22MB (clip 66.47MB, unet 1270.24MB, vae 94.51MB)
[INFO]  stable-diffusion.cpp:2943 - loading model from 'anything-v3-1-ggml-model-q4_0.bin' completed, taking 1.68s
[INFO]  stable-diffusion.cpp:3066 - condition graph use 70.83MB of memory: params 66.47MB, runtime 4.36MB (static 1.43MB, dynamic 2.93MB)
[INFO]  stable-diffusion.cpp:3066 - condition graph use 70.83MB of memory: params 66.47MB, runtime 4.36MB (static 1.43MB, dynamic 2.93MB)
[INFO]  stable-diffusion.cpp:3552 - get_learned_condition completed, taking 0.64s
[INFO]  stable-diffusion.cpp:3568 - start sampling
[INFO]  stable-diffusion.cpp:3260 - step 1 sampling completed, taking 13.10s
[INFO]  stable-diffusion.cpp:3260 - step 2 sampling completed, taking 13.11s
[INFO]  stable-diffusion.cpp:3260 - step 3 sampling completed, taking 13.14s
[INFO]  stable-diffusion.cpp:3260 - step 4 sampling completed, taking 13.15s
[INFO]  stable-diffusion.cpp:3260 - step 5 sampling completed, taking 13.14s
[INFO]  stable-diffusion.cpp:3260 - step 6 sampling completed, taking 13.14s
[INFO]  stable-diffusion.cpp:3260 - step 7 sampling completed, taking 13.16s
[INFO]  stable-diffusion.cpp:3260 - step 8 sampling completed, taking 13.28s
[INFO]  stable-diffusion.cpp:3260 - step 9 sampling completed, taking 13.04s
[INFO]  stable-diffusion.cpp:3260 - step 10 sampling completed, taking 13.04s
[INFO]  stable-diffusion.cpp:3287 - diffusion graph use 1316.63MB of memory: params 1270.24MB, runtime 46.39MB (static 14.67MB, dynamic 31.72MB)
[INFO]  stable-diffusion.cpp:3573 - sampling completed, taking 131.30s
[INFO]  stable-diffusion.cpp:3496 - vae graph use 503.63MB of memory: params 94.51MB, runtime 409.12MB (static 217.12MB, dynamic 192.00MB)
[INFO]  stable-diffusion.cpp:3586 - decode_first_stage completed, taking 32.33s
[INFO]  stable-diffusion.cpp:3600 - txt2img completed in 164.28s, use 1435.58MB of memory: peak params memory 1431.22MB, peak runtime memory 409.12MB
~/stable-diffusion.cpp $

[2haloes]

It works great now, thank you very much!

I've only tested with a tiny image size so far (it would take roughly 20 minutes to generate an image at 512 by 512)