Henry Kroll III
hellork
AI & ML interests
The best models that work for my setup are any of Qwen/Qwen3.6-35B-A3B GGUF with APEX quants.
Consumer-grade (8GB GPU) text generation, image manipulation recognition, 3D model generation, language preservation and translation, and business use cases.
Recent Activity
liked a model about 4 hours ago
sentence-transformers/all-MiniLM-L6-v2 new activity 12 days ago
Jackrong/Qwopus3.5-9B-Coder-GGUF:Reasoning loop in llamacpp new activity 16 days ago
unsloth/Qwen3.6-35B-A3B-MTP-GGUF:5060Ti + Qwen3.6-35B-A3B-MXFP4_MOEOrganizations
Reasoning loop in llamacpp
11
#2 opened 22 days ago
by 
Grandys
5060Ti + Qwen3.6-35B-A3B-MXFP4_MOE
1
#20 opened 19 days ago
by 
dachen125
Absolutely no any speed increase: what wrong with my settings?
👀 1
5
#18 opened 21 days ago
by 
K-Tul-Khu
the trade off is not good (new update)
👍 1
4
#28 opened 22 days ago
by 
rosspanda0
Official llama.cpp slower than the https://github.com/am17an/llama.cpp.git
1
#30 opened 21 days ago
by 
adr-insc
New activity in KevinJK51/Qwen3.6-12B-IQ-Ultra-Heretic-Uncensored-Thinking-V2-Hightop-GGUF 19 days ago
IQ3XXS giving garbage output
2
#33 opened 20 days ago
by 
mbhagya
Official llama.cpp now has '--spec-type draft-mtp'
#29 opened 21 days ago
by hellork
Low VRAM<=8GB best result (40t/s WITHOUT MTP)
8
#17 opened 24 days ago
by hellork
Not working
1
#1 opened about 1 month ago
by 
bgeneto
Just wanted to say thanks Calcuis
😎👍 2
39
#1 opened about 1 year ago
by 
Winnougan
Completely incoherent model when converted Gemma 2 to IQ3_M with Imatrix file
4
#107 opened almost 2 years ago
by 
Dampfinchen
prompt template?
3
#1 opened almost 2 years ago
by 
mirek190
Cannot access gated repo
10
#2 opened about 2 years ago
by 
NickyNicky