Qwen/Qwen3.5-27B · Discussions

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

1

#44 opened 4 days ago by

latent-variable

Qwen 3.5 27b fine tuning multi-gpu example lora sft

#43 opened 10 days ago by

Ethermich

Heavy Chinese-English mixing

#42 opened 11 days ago by

ariesaa

65454

#41 opened 17 days ago by

yehuashangui

what the best engine to run this model.

👀 1

1

#40 opened 19 days ago by

Shimon324

qwen3.5-27b，快速模式下，发票的年份识别不确认，2026识别成2025，在思考模式下才正常

#39 opened 21 days ago by

htl258

Add ScreenSpot-Pro evaluation result

#38 opened 25 days ago by

merve

what's the matrics of livecode bench v6 when enable think off

#37 opened 26 days ago by

HoldenShen

repeat_penalty not repetition

1

#36 opened 26 days ago by

pants44

Add MathArena evaluation result for hmmt/hmmt_feb_2026

#35 opened 26 days ago by

JasperDekoninck

Add MathArena evaluation result for aime/aime_2026

#34 opened 26 days ago by

JasperDekoninck

`.cw` is output at the end of the reasoning content

8

#33 opened 28 days ago by

owao

Chinese-English Typography Issues

#32 opened about 1 month ago by

douglarek

Looping forever

1

#31 opened about 1 month ago by

kil3r

[transformers-server] Continuous batching does not support Qwen-3.5 series

#30 opened about 1 month ago by

zhang-zhi-chun

Language Request: Garhwali

#29 opened about 1 month ago by

akshitgaur2005

What is the chat template for Ollama ?

👍 1

2

#28 opened about 1 month ago by

ngbruce

tutor

#27 opened about 1 month ago by

bububaby

Fine tune with lora

1

#26 opened about 1 month ago by

saireddy

Please don't forget to release the base model of Qwen3.5-27B

👀➕ 10

2

#25 opened about 1 month ago by

Linksome

pure transformers inference code is needed

👀👍 4

3

#24 opened about 1 month ago by

maltoseflower

When?

👀 3

#23 opened about 1 month ago by

AxionLab-official

as a starter

1

#22 opened about 1 month ago by

kit17

Upload 3 files

#21 opened about 1 month ago by

shamamanaeem03

model overthinking

👍 3

1

#20 opened about 1 month ago by

cse2011

vLLM Version conflict issue

4

#19 opened about 1 month ago by

innosynth

有没有BFCLV4和tau2的细节指标

#18 opened about 1 month ago by

HuanchangLiu

Thinking Blocks Error

12

#16 opened about 2 months ago by

TronixAT

Plan on releasing the 4bits quantized version

👀👍 12

#15 opened about 2 months ago by

omarsou

Qwen are is the 8b coming out?

4

#14 opened about 2 months ago by

Crownelius

vllm serving issue

3

#13 opened about 2 months ago by

grozatech

Russian language support, bad grammar!

13

#12 opened about 2 months ago by

alexcardo

Add evaluation results

#11 opened about 2 months ago by

SaylorTwift

Availability of the Qwen3.5-27B Base model

👀👍 16

1

#9 opened about 2 months ago by

Jellyfish042

Value error, Model architectures ['Qwen3_5ForConditionalGeneration'] are not supported for now. Transformers version 5.3.0.dev0

👀 15

6

#8 opened about 2 months ago by

rameshch

Installation Video and Testing - Step by Step

🔥 3

#6 opened about 2 months ago by

fahdmirzac

amazing model

❤️ 7

2

#5 opened about 2 months ago by

Tugay31

is HLE w/ CoT a typo?

#4 opened about 2 months ago by

xiaoqianWX

Samplers

4

#3 opened about 2 months ago by

ggnoy

DENSE

🔥 5

2

#2 opened about 2 months ago by

ox-ox

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 :-)

🚀🧠 3

#1 opened about 2 months ago by

Hanswalter