Jan Reges

janreges3's activity

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

New activity in unsloth/Qwen3.6-27B-NVFP4 2 months ago

NVFP4 vs FP8 throughput on an RTX 6000 Pro 96 GB (Blackwell) - real vLLM numbers

🚀👍 3

2

#9 opened 2 months ago by

janreges3

New activity in nvidia/Qwen3.6-27B-NVFP4 3 months ago

NVFP4 vs FP8 on an RTX PRO 6000 Blackwell (vLLM 0.24): sadly, FP4 is still not faster than FP8 - Qwen3.6‑27B

👀👍 13

1

#8 opened 3 months ago by

janreges3

New activity in rico03/Qwen3.6-27B-Claude-Opus-Reasoning-Distilled 5 months ago

Chat-template fix: occasional </think> leak + body duplication on long-context tasks (patched template attached)

1

#3 opened 5 months ago by

janreges3

Thanks and request for FP8 version

3

#2 opened 5 months ago by

janreges3

New activity in Qwen/Qwen3.5-35B-A3B 7 months ago

vLLM - Looping prevention

👍 1

1

#39 opened 7 months ago by

janreges3

New activity in Qwen/Qwen3-Embedding-4B 10 months ago

How to set MRL (variable dimensions) in vLLM

❤️ 1

5

#21 opened 10 months ago by

janreges3

New activity in BCCard/Qwen3-235B-A22B-Thinking-2507-NVFP4A16 about 1 year ago

Request for NVFP4A16 model Qwen3-30B-A3B 2507 (Thinking & Instruct)

#1 opened about 1 year ago by

janreges3