TonoKen3
sakamakismile
AI & ML interests
Lna-Lab inc.
Recent Activity
liked a model 8 minutes ago
madeby561/GLM-5.2-MXFP8-NVFP4-NF3-Hybrid updated a model 2 days ago
sakamakismile/fable-traces-NVFP4 published a model 2 days ago
sakamakismile/fable-traces-NVFP4Organizations
None yet
Routed expert tensors are half-width — model fails to load (likely a botched export)
1
#5 opened 11 days ago
by
sakamakismile
Good work
1
#1 opened about 1 month ago
by
chris-hc
Thanks
❤️ 1
2
#1 opened about 2 months ago
by
helixdouble
Broken chat_template tool calling and thinking
14
#9 opened 2 months ago
by
livepeer-ren
[Suggestion] Consider AEON-7's BF16 base as an alternative to huihui-ai
4
#1 opened 2 months ago
by
bemoons
SM120 (RTX 5090D) field report + KVTC container availability
1
#1 opened 2 months ago
by
bemoons
Weirdly no perf gain
5
#1 opened 2 months ago
by
Orosius
Docker image lna-lab/gemma4-inference:latest does not exist
1
#1 opened 2 months ago
by
Vossk123
--enable-prefix-caching and Feedback from HGX B200 Deployment
2
#5 opened 2 months ago
by
cgelias
Home Lab Recipe on Blackwell GPU
1
#10 opened 2 months ago
by
SyferO
Does it support tool calling?
2
#4 opened 2 months ago
by
xing120226
RTX RTX PRO 4500 Blackwell results
1
#2 opened 2 months ago
by
Pulsate1680
abliterated?
2
#8 opened 2 months ago
by
celikburak
MTP not responding
9
#7 opened 2 months ago
by
joelafrite
claude code
2
#6 opened 2 months ago
by
chrisqianz
Qwen3.6-27B-NVFP4 is slower than official FP8 on Blackwell; possible fallback / FLA path mismatch on vLLM
👀 2
8
#5 opened 2 months ago
by
garrussun
sglang
1
#4 opened 2 months ago
by
livepeer-ren
vllm部署有问题啊
3
#3 opened 2 months ago
by
xing120226
HOLY $#!T...
❤️🚀 4
2
#1 opened 2 months ago
by
Bellesteck