Hi,

I dont know how to use a forum anymore, been a long time...

Step 1: acquire harware...

I have made a list of hardware I would like to use. I'm going to try to keep on track for it but largely things are going to change due to price fluctuations as well as looking at local markets (FB market place, etc). I'm going to try to host Qwen3.5-35B_A3B on nobara.

Qwen3.5-35B_A3B specs per grok:

Total parameters: 35 billion.
Activated parameters per token: ~3 billion (hence the "A3B" suffix). This comes from a sparse MoE setup with 256 experts total, of which only 8 routed + 1 shared expert activate per forward pass (roughly 3.5% activation rate).
Layers: 40 total.
Hidden dimension: 2048.
Vocabulary size: 248,320 (padded).
Hybrid layout (repeated 10 times): 3 × (Gated DeltaNet → MoE) followed by 1 × (Gated Attention → MoE). This mixes linear attention (for efficiency) with occasional full softmax attention.

Gated DeltaNet details (linear attention component):32 heads for V, 16 for QK.
Head dimension: 128.

Gated Attention details (the "full" attention layers):16 heads for Q, 2 for KV (grouped-query style for efficiency).
Head dimension: 256.
Rotary Position Embedding (RoPE) dimension: 64.

MoE details:Expert intermediate dimension: 512.
This hybrid design (Gated DeltaNet for most layers + sparse MoE FFN blocks) enables high throughput with low latency and minimal KV-cache overhead compared to pure Transformer models.

huggingface.co

The result is a model that behaves like a much denser/larger one during inference but only "pays" for ~3B active parameters. It often outperforms the prior flagship Qwen3-235B-A22B (dense or MoE variants) on many tasks while being dramatically cheaper and faster to run.Multimodal CapabilitiesIt is a unified vision-language model trained with early-fusion on multimodal tokens. It natively supports:Text.
Images.
Video.

Strengths include:Visual reasoning.
Spatial grounding.
Document/UI analysis.
OCR.
Image captioning.
GUI interaction.

It outperforms the previous Qwen3-VL series in these areas and achieves cross-generational parity (or better) with larger Qwen3 models on multimodal benchmarks. It handles tool use and agentic workflows that combine vision + reasoning + code.

lmstudio.ai

Context and EfficiencyNative context length: 262,144 tokens (~393 A4 pages of text).
Extendable to ~1 million tokens via RoPE/YARN scaling.
Strong context scaling: throughput degrades very little even at long contexts (e.g., only ~0.9% drop in some tests from 512 to 8k+ tokens).
Supports up to ~65k output tokens in many deployments.

Inference speed is a major selling point:On consumer hardware (e.g., quantized Q4/Q5 or FP8):RTX 3090 (~$800 used): ~110–112 tokens/second at full 262k context.
RTX 4090: ~122 tokens/second.
RTX 5090: up to ~170 tokens/second.
Apple Silicon (M1/M4 Macs, 24–64GB unified memory): 15–37+ tokens/second depending on quantization and prompt length.
High-end servers (H100/H200): hundreds to over 1,000 tokens/second in batched/optimized setups.

Quantized versions (e.g., 4-bit, FP8, GGUF) fit comfortably on 24GB VRAM cards or even lower with offloading. Some reports show it running on as little as 8–19GB VRAM effectively.

agentnativedev.medium.com

It is designed for production: low latency, high throughput, and suitable for local/agentic use cases (many users report it replacing multi-model setups for coding + reasoning workflows).Performance HighlightsThe model shines in efficiency-vs-performance tradeoffs. It delivers:Reasoning and knowledge: Competitive with or better than much larger models. Examples from benchmarks:MMLU-Pro: ~85.3%.
GPQA Diamond: ~84.2%.
Strong on math (e.g., HMMT), multilingual (C-Eval), and general knowledge.

Coding and agentic tasks: Excellent at software engineering, tool use, and multi-step agentic workflows (planning + coding + analysis). Users frequently call it a "gamechanger" for local coding agents, often handling full tasks (data analysis + code + insights) in a single model where previously two were needed.
Multimodal: Improved over prior VL models in document understanding, visual reasoning, and GUI tasks.
Overall intelligence indices place it high among open models of its class, sometimes approaching or rivaling closed models like Claude Sonnet variants or GPT-5-mini in specific domains (reasoning, coding, agents).

It supports 201 languages and dialects with nuanced cultural understanding. Post-training includes scaled reinforcement learning (RL) across complex, progressively harder tasks for better real-world generalization.Other Practical DetailsLicense: Apache 2.0 (fully open weights; base and instruction-tuned variants available on Hugging Face).
Deployment: Works with Hugging Face Transformers, vLLM, llama.cpp, Ollama, LM Studio, MLX (Apple), etc. FP8 and various GGUF quants are popular.
Hosted versions: Powers things like Qwen3.5-Flash on Alibaba Cloud (fast/cheap tier).
Uncensored variants: Community fine-tunes exist for minimal refusals.
Limitations (as with any model): Still has occasional hallucinations on edge cases; multimodal performance depends on input quality; very long contexts can still tax lower-end hardware in prefill phase.

In short, Qwen3.5-35B-A3B is one of the best examples yet of how architectural innovations (hybrid linear attention + sparse MoE + high-quality training/RL) can break traditional scaling expectations. It gives frontier-like capabilities (reasoning, coding, vision, agents) at the inference cost of a much smaller dense model, making it highly practical for local runs, agentic systems, and cost-sensitive production use. Many in the community view it as a turning point for accessible high-intelligence local AI.

https://www.youtube.com/watch?v=2-mTJHheqXo

https://www.youtube.com/shorts/t77d9cIJhx4

A plugin that allows you to stream 3D View from Blender into Krita, behaving like a normal layer: You can trace over it, change the blending mode and even apply layer styles!

Krita has so much cool plugins I'd honestly replace most programs with it.
Fuck adobe btw shit runs like ass.

Maybe make tutorials our various linux experiences? Would ya'll be down for that?

The odd scoop shaped penis of primates has been the object of fascination by biologists for ages. Why is it that the head of the penis is shaped the way it is?

A background in cuckoldry

Turns out that cuckoldry has been a part of human evolution for as long as humans have existed. Two of the top ten reported sexual fantasies of females is having sex with multiple males at the same time, and having sex with males other than their partner.

This act of double-mating appears to be a real outcome as well. 17.5% of females that have had fewer than 50 copulations reported that they had engaged in double-mating at least once. The figure seems to increase steeply with sexual experience. 71.8% of females that have had more than a 1000 copulations, also reported having double-mated at least once.

Naturally, some might say, this leads to a lot of offspring genetically unrelated to their social father. Current estimates are that 3.1% of children in Western countries are genetically unrelated to their social father.

The tendency for females to double-mate has led to the need for males to adapt and develop sperm competition tactics, and anti-cuckoldry tactics.

Semen displacement theory

Current evidence suggests that one of the anti-cuckoldry tactics is the ability to displace semen within the female's reproductive tract with their own. Using artificial human penises (realistic dildos essentially), and artificial reproductive tracts (fleshlights), scientists were able to provide evidence that the human penis may be able to displace semen within the female. Combining the overall scoop-shape and deep thrusting, the scoop-shaped penises could remove a rivals sperm far more efficiently than non-scoop-shaped penises.

The results seems staggering. Penises with a mushroom-like appearance displaced upwards of 90% of the sperm already within the artificial female, as opposed to only around 30% in the smooth penis. Being able to scrape out a rival male's sperm and replace it with your own seems to be one of the significant anti-cuckoldry and sperm competition tactics employed by males.

Sources

DeLecce, T. (2022). Sperm Competition Theory. In T. K. Shackelford (Ed.), The Cambridge Handbook of Evolutionary Perspectives on Sexual Psychology (pp. 242–263). chapter, Cambridge: Cambridge University Press. https://doi.org/10.1017/9781108943567

Gallup, G. (2003). The human penis as a semen displacement device. Evolution and Human Behavior, 24(4), 277–289. https://doi.org/10.1016/s1090-5138(03)00016-3

https://www.youtube.com/watch?v=N7FgIKaBM3g&list=PLACumMv6MnNcVs3gFVfpuiuRlUpJMVaau

For those who are curious; Red onions are formally named Allium Cepa. The photo shows completely turgid (full of water and rigid) epidermal (outer-most layer) cells under 40x light microscopy magnification.

Fun fact! Plants stay firm and rigid by having their cells almost burst by "overfilling" them with water. They do not burst like animal cells due to their thick cell walls. Kind of like how water balloons gets firmer the more you fill them. Your own blood cells would likely burst and die under the same conditions.

https://www.youtube.com/watch?v=9ELpTHMamGE

![2252.jpg]

Bazzite does bazzite things

Welcome to your brand new NodeBB forum!

This is what a topic and post looks like. As an administrator, you can edit the post's title and content.
To customise your forum, go to the Administrator Control Panel. You can modify all aspects of your forum there, including installation of third-party plugins.

Additional Resources

• NodeBB Documentation
• Community Support Forum
• Project repository