John
cmp-nct
AI & ML interests
I am a ML engineer with two decades of experience in data processing and AI.
Demodokos Foundry is the most advanced local Speech and Music AI Studio.
Version 2.0 has been released supporting the updated V4 speech model - being right on eye level with the best you can find in the cloud today.
Version 2.1 is under active development as a great stop forward for professional local music generation
Recent Activity
updated a model about 13 hours ago
cmp-nct/demodokos-foundry-tts-v4 updated a model about 16 hours ago
cmp-nct/demodokos-foundry-music-v4 liked a model 5 days ago
cmp-nct/demodokos-foundry-music-v4Organizations
The first small model for honest conversation
👍 7
4
#22 opened 3 months ago
by 
ghostwithahat
llama.cpp inference - 20 times (!) slower than OSS 20 on a RTX 5090
➕ 1
9
#12 opened 6 months ago
by cmp-nct
Interesting model - how to control reasoning length?
#2 opened 6 months ago
by cmp-nct
Should UD-Q6_K_XL identical to Q6_K.gguf?
5
#1 opened 7 months ago
by 
BVEsun
Quantizations below 4 bit are bad
1
#4 opened 7 months ago
by cmp-nct
Good model but it is very flawed in recalling input
6
#5 opened 7 months ago
by cmp-nct
No space to test it, not on open router, are there obstacles ?
2
#9 opened 8 months ago
by cmp-nct
Kernel assertion errors on 5090 using generation with MXfp4 (gpt-oss) - (stable on 4090)
5
#1 opened 11 months ago
by cmp-nct
Creating quantization
#4 opened 11 months ago
by cmp-nct
This model is not much better than qwen3 32b for writing code
👍 3
9
#4 opened 12 months ago
by 
xldistance
Wan't to share my Training-Set
3
#4 opened about 1 year ago
by 
havok2
Very good results
3
#5 opened about 1 year ago
by cmp-nct
Multiple languages support
👀👍 15
6
#19 opened about 1 year ago
by 
thunder-007
Very coherent for the size, though it is also adamant in defending its wrong conclusions and false knowledge
👍 1
1
#5 opened about 1 year ago
by cmp-nct
Interesting idea, what about multi lingual prompts ?
➕ 3
3
#7 opened about 1 year ago
by cmp-nct
Gguf?
👍 8
8
#2 opened over 1 year ago
by 
AlgorithmicKing
Safety
👍 1
6
#21 opened over 1 year ago
by 
Javier81k
Modl showing up as Llama instead of phi3 in LMstudio
12
#1 opened about 2 years ago
by 
Tirin
gguf / llama.cpp support
2
#45 opened about 2 years ago
by cmp-nct