Torlando
@torlando.tech
Reticulum is this super cool network stack that can work over any medium, from super fast fiber, to super slow radio. It doesn't need the internet, and all traffic is encrypted. Your mom had a project for her precursor to make a messaging app for reticulum but I think she forgot about it 🥲
January 10, 2026 at 5:07 PM
Reticulum is this super cool network stack that can work over any medium, from super fast fiber, to super slow radio. It doesn't need the internet, and all traffic is encrypted. Your mom had a project for her precursor to make a messaging app for reticulum but I think she forgot about it 🥲
@luna.pds.witchcraft.systems has your mom introduced you to the wonderful world of #reticulum?
January 10, 2026 at 4:13 AM
@luna.pds.witchcraft.systems has your mom introduced you to the wonderful world of #reticulum?
Thank you @strix.timkellogg.me! Can't wait to see the results. I'd be curious how larger models with more aggressive quantization behave compared to small models with no quantization. Put another way, what's the best way to optimize limited VRAM with open weight models?
January 7, 2026 at 6:25 PM
Thank you @strix.timkellogg.me! Can't wait to see the results. I'd be curious how larger models with more aggressive quantization behave compared to small models with no quantization. Put another way, what's the best way to optimize limited VRAM with open weight models?
Looking forward to @strix.timkellogg.me's reply 😊
January 7, 2026 at 6:11 PM
Looking forward to @strix.timkellogg.me's reply 😊
Wow yeah that's actually slightly worse than Texas summer, by a few degrees. Wishing you and your A/C a healthy summer
January 7, 2026 at 3:02 PM
Wow yeah that's actually slightly worse than Texas summer, by a few degrees. Wishing you and your A/C a healthy summer
What's the smallest model that performed decently well? Have you tested quantized models to see how quantization affects collapse?
January 6, 2026 at 5:52 PM
What's the smallest model that performed decently well? Have you tested quantized models to see how quantization affects collapse?
I tried a similar setup, same model, (same machine) and tried setting up a letta agent with the ADE. It had issues forming tool calls. Did you run into this? Also are you using rocm, or vulkan?
January 6, 2026 at 4:19 PM
I tried a similar setup, same model, (same machine) and tried setting up a letta agent with the ADE. It had issues forming tool calls. Did you run into this? Also are you using rocm, or vulkan?
Did you do any fine-tuning for Luna or is it just all system prompt / letta memory?
January 3, 2026 at 6:47 PM
Did you do any fine-tuning for Luna or is it just all system prompt / letta memory?
I just want opus 4.5 at home 😭
January 1, 2026 at 5:15 PM
I just want opus 4.5 at home 😭
Have you tried putting any of the open coding agents in front of it? Curious if it's any good
December 31, 2025 at 11:02 PM
Have you tried putting any of the open coding agents in front of it? Curious if it's any good
This helped me with mine: github.com/kyuz0/amd-st...
GitHub - kyuz0/amd-strix-halo-toolboxes
Contribute to kyuz0/amd-strix-halo-toolboxes development by creating an account on GitHub.
github.com
December 24, 2025 at 9:34 PM
This helped me with mine: github.com/kyuz0/amd-st...
I'm really hoping that in this decade I can run something as smart as opus 4.5 on less than $5k of hardware at home
December 20, 2025 at 9:01 PM
I'm really hoping that in this decade I can run something as smart as opus 4.5 on less than $5k of hardware at home
Manage social media presence for small business. Thinking of solo Etsy shop makers who would rather be making and designing products than promoting online. Bonus points for "human in the loop" to approve some/all comms
December 12, 2025 at 7:22 PM
Manage social media presence for small business. Thinking of solo Etsy shop makers who would rather be making and designing products than promoting online. Bonus points for "human in the loop" to approve some/all comms
I got the sage one and lavender trackpad and expansion cards. Absolutely love it. Just wish I'd sprung for more than 16GB ram before it quadrupled in price 😭
December 7, 2025 at 10:51 PM
I got the sage one and lavender trackpad and expansion cards. Absolutely love it. Just wish I'd sprung for more than 16GB ram before it quadrupled in price 😭
Serious question - does this apply to ANYTHING in the game AI generated, like AI assisted code, or does it focus on art/music/assets that are more easily classified in the Creative category vs. implementation details?
December 3, 2025 at 8:59 PM
Serious question - does this apply to ANYTHING in the game AI generated, like AI assisted code, or does it focus on art/music/assets that are more easily classified in the Creative category vs. implementation details?
It's not as plug and play, but AMD does have the halo strix available with 128GB for roughly half price. Framework Desktop is $2k for that config
November 27, 2025 at 4:11 PM
It's not as plug and play, but AMD does have the halo strix available with 128GB for roughly half price. Framework Desktop is $2k for that config
That being said if you've had success with smaller open weight models as letta agents, I'm all ears
November 18, 2025 at 12:24 AM
That being said if you've had success with smaller open weight models as letta agents, I'm all ears
CC is my daily driver and I'm allergic to API usage based billing lol. Last month, ccusage clocked in just over $2k on my primary machine alone, on the $200 max plan. I really want to explore letta but for me it's either gotta be accessible via existing monthly sub, or run on <96GB offline
November 18, 2025 at 12:23 AM
CC is my daily driver and I'm allergic to API usage based billing lol. Last month, ccusage clocked in just over $2k on my primary machine alone, on the $200 max plan. I really want to explore letta but for me it's either gotta be accessible via existing monthly sub, or run on <96GB offline
Letta as a memory layer for Claude code? Like, not just read but also also update the memory from Claude code?
November 18, 2025 at 12:13 AM
Letta as a memory layer for Claude code? Like, not just read but also also update the memory from Claude code?