Recent changes

Jump to navigation Jump to search

Track the most recent changes to the wiki on this page.

Recent changes options Show last 50 | 100 | 250 | 500 changes in last 1 | 3 | 7 | 14 | 30 days
Hide registered users | Hide anonymous users | Hide my edits | Show bots | Hide minor edits
Show new changes starting from 16:07, 29 August 2026
   
List of abbreviations:
N
This edit created a new page (also see list of new pages)
m
This is a minor edit
b
This edit was performed by a bot
(±123)
The page size changed by this number of bytes

27 August 2026

N    09:20  Category:Bonsai diffhist +40 PeterHarding talk contribs Created page with "Bonsai ML links, notes and references..."
N    09:19  Using Bonsai QWEN Models‎‎ 2 changes history +2,284 [PeterHarding‎ (2×)]
     
09:19 (cur | prev) +23 PeterHarding talk contribs
N    
09:18 (cur | prev) +2,261 PeterHarding talk contribs Created page with " See - https://www.youtube.com/watch?v=V6LmF7TuBmY (Bonsai 27B Runs Qwen 3.6 27B at 10x less memory) Not a corrupt download — a format mismatch. Q2_0 in that filename isn't stock llama.cpp's quant; it's Q2_0_g128, PrismML's custom ternary packing (2-bit slots, one FP16 scale per 128 weights) that only their fork's kernels understand. Your stock build computes a different byte size for the ternary tensors, so its running offset drifts from what's written in the header,..."
N    09:19  Category:Llama.cpp diffhist +40 PeterHarding talk contribs Created page with "llama.cpp links, notes and references..."
N    09:19  AI Notes‎‎ 3 changes history +131 [PeterHarding‎ (3×)]
     
09:19 (cur | prev) −2,177 PeterHarding talk contribs Replaced content with "= A Collection of AI Related Notes = * Using Bonsai QWEN Models Category:AI Category:QWEN Category:Llama.cpp" Tag: Replaced
     
09:17 (cur | prev) +32 PeterHarding talk contribs
N    
09:16 (cur | prev) +2,276 PeterHarding talk contribs Created page with "= A Collection of AI Related Notes = See - https://www.youtube.com/watch?v=V6LmF7TuBmY (Bonsai 27B Runs Qwen 3.6 27B at 10x less memory) Not a corrupt download — a format mismatch. Q2_0 in that filename isn't stock llama.cpp's quant; it's Q2_0_g128, PrismML's custom ternary packing (2-bit slots, one FP16 scale per 128 weights) that only their fork's kernels understand. Your stock build computes a different byte size for the ternary tensors, so its running offset drif..."