froggeric

froggeric's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

New activity in froggeric/Qwen-Fixed-Chat-Templates 6 days ago

inference throughput drops by 80% with this template

14

#28 opened 10 days ago by

froilo

New activity in froggeric/Qwen-Fixed-Chat-Templates 13 days ago

v19 released with major improvements

❤️🚀 17

22

#22 opened 15 days ago by

froggeric

New activity in froggeric/Qwen3.6-27B-MTP-GGUF 14 days ago

error while handling argument "--spec-type": unknown speculative type: mtp

1

#6 opened 17 days ago by

felkf

New activity in froggeric/Qwen-Fixed-Chat-Templates 15 days ago

`preserve_thinking` flag ignored?

2

#9 opened 21 days ago by

astride-thee-squid

v16 failes to parse JSON correctly - Tool Call error

👍 1

2

#20 opened 16 days ago by

herstrabol

v18 broken tool calling

5

#21 opened 15 days ago by

slepkaviba

final v16 does not appear to work correctly, it stops after the first prompt.

🔥 2

2

#19 opened 17 days ago by

mancub

New activity in froggeric/Qwen-Fixed-Chat-Templates 17 days ago

Qwen 3.6 template has unnecessary cache invalidations

14

#1 opened about 1 month ago by

sdroege

vLLM: 'Replace the chat_template string in your tokenizer_config.json with the file contents.' ... where?

4

#5 opened 23 days ago by

pathosethoslogos

When using Claude Code, tool calls end up broken with this chat template in Qwen3.6-27B

7

#6 opened 23 days ago by

mancub

think off , but answer in reasoning

1

#7 opened 22 days ago by

xd2333

Improvement (50% token reduction via the tool)

7

#8 opened 22 days ago by

Elsephire

Template V10 causes hallucinations.

5

#10 opened 21 days ago by

lorenzodimauro97

v11/v12 performance considerations with Claude Code?

3

#11 opened 21 days ago by

mancub

test v12 + qwen 27B + llamacpp + pi coding agent

2

#12 opened 20 days ago by

tooltd

Abruptly stopping with <tool call>

2

#13 opened 20 days ago by

crotron

v13 stops dead after the first response

👍 2

5

#14 opened 20 days ago by

mancub

v15 released with fixes for overthinking/indecision and loops on tool errors

❤️ 3

4

#15 opened 18 days ago by

froggeric

v13 to v15 are completely broken in VLLM for qwen3_coder parser

5

#16 opened 18 days ago by

gudsjwixvfasptjhtr

Python_Search_file breaks the session

1

#17 opened 18 days ago by

Jokeru1983