Michael Meding PRO

mmeding's activity

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

New activity in dealignai/DeepSeek-V4.1-Flash-1017-EXL3-2.9bpw-UNCENSORED 6 days ago

Performance: 54 projections in the 1017 EXL3 packs are stored as BF16 (~24% slower TP decode) — cause and fix

4

#3 opened 6 days ago by

mmeding

liked a model 3 months ago

deepseek-ai/DeepSeek-V4-Flash-0731

Text Generation • 304B • Updated Aug 1 • 2.74M • • 4.1k