Jose Marcalo

JigSawPT's activity

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

New activity in deepseek-ai/DeepSeek-V4.1-Flash about 2 months ago

Runs on one RTX 5090 (31.8 GiB) + 125.7 GiB RAM via a llama.cpp fork: GGUF, report and numbers

🔥 5

4

#37 opened 2 months ago by

JigSawPT

New activity in JigSawPT/DeepSeek-V4.1-Flash-GGUF about 2 months ago

Request of a Q2 quantization in the "antirez" style?

🚀 1

1

#3 opened 2 months ago by

automacellulare

Laptop RTX5090 (24GB vRAM) + 64GBRam + RAID0 Gen4 -> 3.45 token/sec

5

#2 opened 2 months ago by

aydintb

I am going try this MXFP4 quant on 4 TITAN X, 12.2, GTX 960 4GB + 330GiB DDR3 + Samsung m.2 NVME for the checkpoint.

3

#4 opened about 2 months ago by

BigDeeper

New activity in JigSawPT/DeepSeek-V4.1-Flash-DSpark-GGUF 2 months ago

DeepSeek-V4.1-Flash Suggestion

3

#1 opened 2 months ago by

TheRealRalphWiggum

Keep the target's token_embd unquantized: +4–5 points of acceptance

1

#2 opened 2 months ago by

midagedev

New activity in JigSawPT/DeepSeek-V4.1-Flash-GGUF 2 months ago

How this GGUF runs on one RTX 5090, and the numbers behind it

👍 2

2

#1 opened 2 months ago by

JigSawPT