peculiar-ragdoll
/

Qwen-Sharp-Chat-Templates

token-efficient

Model card Files Files and versions

Resources

View closed (1)

compatible with Qwen3.8-Flash-Next too???

#12 opened 5 days ago by

Production adoption report (vLLM + MTP): terseness confirmed, quality up at n=4 — and the MTP think-leak did NOT reproduce

#11 opened 10 days ago by

Would this work on Kay coder dev?

#10 opened 12 days ago by

reasoning_effort info is outdated

#9 opened 13 days ago by

Fix KV cache reuse when changing reasoning_effort

#8 opened 16 days ago by

Usage on Ornith-1.5-35B-A3B

#7 opened 17 days ago by

Add a `_default_reasoning_effort` knob and remove the dead `_initial_effort` line (discussion #3)

#6 opened 18 days ago by

Add a `terse` template kwarg to opt out of the terseness block

#5 opened 18 days ago by

Make reasoning effort steering more intuitive in the Jinja template

#3 opened 20 days ago by

Froggeric just release the new Qwen-Fixed-Chat-Templates v22

#2 opened 25 days ago by

Possible enhancemet to avoid problems with tool calling

#1 opened 25 days ago by