Qwen
/

Qwen3.5-397B-A17B-FP8

Image-Text-to-Text

Model card Files Files and versions

Resources

View closed (1)

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

#12 opened 13 days ago by

latent-variable

Missing out <think> token when calling model and usually miss '{' token in arguments field when use tool calling.

#11 opened about 1 month ago by

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 :-)

#10 opened about 2 months ago by

When can we expect 4B, 8B, and 10B versions?

#8 opened 2 months ago by

Is there possible to see 32B dense version?

#7 opened 2 months ago by

Memory Requirements to run `Qwen/Qwen3.5-397B-A17B-FP8`

#6 opened 2 months ago by

Gibberish output with vllm nightly qwen3_5 build

#5 opened 2 months ago by

mxfp4 QAT versions?

#4 opened 2 months ago by

can we deploy this using tp (H100-80GB each)=6 ?

#3 opened 2 months ago by

ValueError: Weight output_partition_size = 16 is not divisible by weight quantization block_n = 128

#2 opened 2 months ago by

Qwhen are you making an 8b version???

#1 opened 2 months ago by