OpenSlop

Open source Slop Management Suite. Use our Dedicated Slop Factory, Distributed Slop Servers, SwarmBots and Agents for meaningless work TODAY.

[OpenSlop @TensorTease]

Try https://t.co/GXzYZxhWsH. https://t.co/0j90JQQZ7I

[OpenSlop @TensorTease]

I updated hoping that would fix the issue of not finding an executable on a GDX10 install. It did not. The bug was in desktop_packaged_executable line 7224: the Linux branch only looked for linux-unpacked/ but electron-builder names the output directory with the arch suffix when targeting a specific architecture — so linux-arm64-unpacked/ was never in the search list.

[OpenSlop @TensorTease]

@NousResearch Attempted install on GX10 machine, failed. Hour wasted looking for human readable documentation, failed. Hermes has suggested not wasting more time on it, smart agent.

[OpenSlop @TensorTease]

I had an AXE16000 router not doing anything so I set up an isolated network for Hermes to use for agent to agent communication as well as handle video feeds. An RTX 3060 running YOLOv8 and Llava:7b handles the capture, analysis and flagging. Hermes polls the Llava log from it's GX10 and alerts me if something needs my attention.

[OpenSlop @TensorTease]

@PyTorch Impressive. Benchmark was on an x86, think it will operate on an ARM64 machine (GB10)?

[OpenSlop @TensorTease]

Trying @NousResearch 's Hermes. So far so good, Hermes and @claudeai Code negotiated areas of responsibility and communications pretty quickly. Comms done thru a handoff file to avoid burning too many tokens. Local TTS only took a prompt or two.

[OpenSlop @TensorTease]

@ClementDelangue @huggingface Added my rigs, 379.17 TFLOPS of slop generating capacity. Who knows the next Medicare ad may have been created on my machines. Don't hate the player.

[OpenSlop @TensorTease]

I'm running @claudeai Code to tune the openclaw/qwen3.6:27b nvfp4. qwen3.6 gets stuck in loops sometimes when updating skills. Claude code is pretty impressive though, /init and off it goes.

[OpenSlop @TensorTease]

I finally gave up on Nvidia's NemoClaw for GX10. I'm running OpenClaw with Qwen3.5:27b. Bootstrapping was easy enough.

[OpenSlop @TensorTease]

Unboxed the Asus Ascent GX10. Could not install nemoclaw. https://t.co/0BGU0kZ1ww — Founders Edition dependency that ASUS never implemented, blocking Docker from starting entirely
Corrupted buildkit database — GX10 first-boot issue, nuked and rebuilt cleanly.

[OpenSlop @TensorTease]

I raised my https://t.co/qVV3DI2r94 rates today. I did not mind being underpriced so long as gpu were not worked too hard. I'm seeing heavy loads so.... I'm still underpriced, we'll see in the adjustment works.

[OpenSlop @TensorTease]

I congratulate @vast_ai for breaking 90% occupancy, a feat I thought not possible. Compute demand outpacing supply.

[OpenSlop @TensorTease]

@traviscannell My machines have been booked 24/7. Adding a new server using RTX cards not cost effective, for me anyhow. The DGX GB10s are about the only systems that make economic sense and I'm picking my first up this week. If you need a sandbox to test Vast in, let me know.

[OpenSlop @TensorTease]

I'm still kicking around ideas for uses of the GB-10 machine. Likely set up an environment for agents to compete in adversarial goals.

Demand and rates for RTX Pro 6000 WS are heading up on Vast ai, I was wondering with the cooling season coming if I would have to raise rates to offset cooling. My current client's moderate use and 24/7 booking leads me to believe I'll extend current rates another month. TY whoever you are.

[OpenSlop @TensorTease]

https://t.co/GXzYZxhWsH: I think 75% occupancy is full utilization. Geography, maintenance, comms all play a role is GPU suitability, so their current 70%+ does not leave much headroom. RTX prices don't support onboarding, DGX & Arc Pro B-70 unsupported. Growth will be difficult

[OpenSlop @TensorTease]

My @vast_ai prices are discounted, but as a host it's good business. My machines are rented 24/7 and the loads on the GPUs modest. GPU temps average 55'. Client wins with 24/7 instant access I win with lighter utility bills, the margins favor both. Picking up a GB10 soon.

[OpenSlop @TensorTease]

My wife's employer want her to train their inhouse AI on her voice. We're seriously considering having her copywrite her voice and sell the rights to me. She would not be able to train the AI without my permission. We'll have to speak with an IP attorney. Sick New World.

[OpenSlop @TensorTease]

Upgraded my UPS today, raised rates on the 5090's, 1 month 24/7 session https://t.co/qVV3DI2r94 @vast_ai , dropped the rates for the Pro 6000 WS. Relaibility took a hit for upgrade. I've non-zero change of verification. Got MIG skills? my 6000 WS is a Multi Model Slop Factory

[OpenSlop @TensorTease]

My slop scraper tossed an OOM error before it even finished with @elonmusk 's X account.

[OpenSlop @TensorTease]

As I look at the landscape today, over 90% of the Models are Slop Factories. I was about to activate the DSS, then it occurred to me HF already provides a centralized SS for DSFs. VC for ODSFs? Not out of the question, Series A then boom, $1B valuation.