Resources

fix chat template to avoid empty historical `<think>` blocks

👍 22

#22 opened 5 days ago by

latent-variable

Add YC-Bench benchmark result (avg $0)

#21 opened 11 days ago by

RiddleHe

PLS BASE ver.

#20 opened 15 days ago by

EvilinaMaller

Model loading reduction from 5.5 minutes to 5.26 seconds

#19 opened 17 days ago by

tcclaviger

MTP Working on AMD 4xR9700 and MXFP4 Custom Kernel

👍 1

#18 opened 19 days ago by

tcclaviger

Add ScreenSpot-Pro evaluation result

#17 opened 26 days ago by

merve

40B to 80B dense LLM: A better/smarter model than 122B that can still fit

#16 opened about 1 month ago by

rtzurtz

Not abliterated

#15 opened about 1 month ago by

Korberich

issue with vllm running

#12 opened about 2 months ago by

aathi1324

qwen3.5本地部署后都是使用英文进行推理，这样对吗？

#11 opened about 2 months ago by

mimeng1990

hope to be able to release nightmedia/Qwen3.5-122B-A10B-Text-qx86-hi-mlx,

#10 opened about 2 months ago by

mimeng1990

Qianwen is amazing! He made so many models of different sizes. This is a wonderful thing for the people!

🤗 1

#9 opened about 2 months ago by

mimeng1990

Add evaluation results

#8 opened about 2 months ago by

SaylorTwift

感谢qwen团队的工作！

#7 opened about 2 months ago by

Simon716

Will there be an Instruct version?

#6 opened about 2 months ago by

a-r-c

Speed is so good especially the prompt preprocessing, much much faster than coder-next

#5 opened about 2 months ago by

rosspanda0

Official FP8

👀👍 14

#4 opened about 2 months ago by

retowyss

Thank you team Qwen for a 120B LLM

❤️ 3

#3 opened about 2 months ago by

rtzurtz

Benchmarks against Qwen Coder Next 80B

👍👀 10

#2 opened about 2 months ago by

ztsvvstz

请千万千万别忘了把 Qwen Image 2.0 也开源——这对我们本地用户来说会是个巨大的改变 :-)

🤗 2

#1 opened about 2 months ago by

Hanswalter