JM
akierum
·
AI & ML interests
None yet
Recent Activity
new activity about 1 hour ago
Qwen/Qwen3.8-27B:DFlash draft model instead of MTP one new activity about 1 hour ago
Qwen/Qwen3.8-27B:It is inferior to the qwen3.6 35 model in image and video recognition new activity about 1 hour ago
Qwen/Qwen3.8-27B:Why Qwen3.8-27B overthinks? Here the reason.Organizations
None yet
DFlash draft model instead of MTP one
👍 1
8
#54 opened 2 days ago
by
artden111
It is inferior to the qwen3.6 35 model in image and video recognition
3
#69 opened 2 days ago
by
wzgrx
Why Qwen3.8-27B overthinks? Here the reason.
❤️👀 24
24
#76 opened 1 day ago
by
LuffyTheFox
codeneedle benchmark results
2
#33 opened 2 days ago
by
akierum
Performance report on RTX 5090: 100 t/s with UD-Q6_K_XL
7
#14 opened 2 days ago
by
SlavikF
建议延续马云的思想将千问团队剥离出来独立发展,借助资本做大做强
👍 1
1
#23 opened 3 days ago
by
emei8
122b a10b?
➕👍 18
6
#17 opened 4 days ago
by
TheBigBlockPC
27B where
🔥🚀 14
9
#4 opened 4 days ago
by
pengzhi27
DSpark MTP
5
#1 opened 6 days ago
by
cpuq
Long-context performance beyond 32K may require RoPE/YaRN scaling.
#6 opened 6 days ago
by
akierum
Why no GGUF release?
#3 opened 16 days ago
by
akierum
What are best llamma.cpp run settings for opencode etc.?
4
#9 opened 22 days ago
by
akierum
MTP support? Why no Vision?
20
#3 opened 24 days ago
by
unoid
Pretty good
👍 2
7
#2 opened 27 days ago
by
flyingweasel
No Vision where is mmproj file
#3 opened 22 days ago
by
akierum
Mini Model?
2
#1 opened 23 days ago
by
Akicou
llama tool calling
👍🤗 3
4
#2 opened 23 days ago
by
xenarathon