TechnoRollSushiTroll

Dad, husband, dog lover. Passions for engineering solutions, quantum computing, neural networks, and Blockchain technology. A lover of music.

[TechnoRollSushiTroll @TL_Benson]

@davepl1968 No fair just repeating what the docs said!

[TechnoRollSushiTroll @TL_Benson]

@KTVUtraffic Sal what happened! Santa Rosa to SF way over 2 hours. Novato101 a parking lot. Ave exit to Ignacio blvd on-ramp on city streets, 45 minutes. Over 90 minutes driving & still over 90 minutes left to the Mission!? I made it home from Ignacio in under 25 minutes...

[TechnoRollSushiTroll @TL_Benson]

@theo If they were to do a face-off how do you think they would compare?
- Spec based planning
- Asking pertinent questions
- Sub agents / Plan Implementation
- CI/CD GitHub Actions, GitLab Pipelines (or Atlassian Bamboo)
- Terraform
- Python
- Web HTML/CSS?

[TechnoRollSushiTroll @TL_Benson]

@davepl1968 With "same transceiver on both ends" is this a system to system w/o any switch in between? Any more context you can share? Like test method/software/tools being used? What workload or simulated workload in each direction? "at all defaults" I presume MTU 1500 on both sides?

[TechnoRollSushiTroll @TL_Benson]

@Nelsondsa19 @NotebookLM I've not experienced choppy audio for the podcast, or the video, in literally dozens of notebooks.

If you find that audio is choppy pretty consistently I suggest checking your system load. See if there are any CPU spikes that could lead to poor audio quality.

[TechnoRollSushiTroll @TL_Benson]

I've been enjoying using @NotebookLM lately. The simplicity of the video breakdown vs. the podcasts deep dive. To be able to interact with the podcast hosts is hillarious, but useful too. Join to interrupt the flow and ask your question when it comes to mind, priceless.

[TechnoRollSushiTroll @TL_Benson]

@awsdevelopers AWS

[TechnoRollSushiTroll @TL_Benson]

@theo 32.2K pushed you over the limit? That's a ton of prior context. You aren't locking the model to Haiku are you?

[TechnoRollSushiTroll @TL_Benson]

If you try out any of these Notebooks let me know what you think about the interactive mode in the podcast. While it might not be everyone's cup of tea, I'm sure some are going to find it interesting.

[TechnoRollSushiTroll @TL_Benson]

And one last one where Opus 4.6 and GPT 5.3 releases are reviewed: https://t.co/SKdukKObho

[TechnoRollSushiTroll @TL_Benson]

@grok @OpenAI @opencode @ollama @vllm_project So far its been Claude Code. Gemini has been impressive though. That superagent-ai/grok-cli is the exact one I was thinking about. I'll soon need a full on matrix of which CLI tool and which model it used though!

@grok Please build that test matrix for me and save me the time!

[TechnoRollSushiTroll @TL_Benson]

Foundational models are getting pretty impressive. I play with just about every new tech, AI or not. Whether it's for work or personal projects, I love to spin up an environment. Recently I've been playing with both @AnthropicAI's Claude Code and @GoogleAI's Gemini CLI.

[TechnoRollSushiTroll @TL_Benson]

0.28.1

[TechnoRollSushiTroll @TL_Benson]

I asked Gemini CLI to use its Conductor w/ Gemini 3 to setup the agent harness. Better than Gemini 2.5, check. Then I realized that if I asked to hand off a task to a sub-agent, it flawlessly used the sub-agent instead of run directly, further bloating the context window👍

[TechnoRollSushiTroll @TL_Benson]

2.1.39 (Claude Code)

[TechnoRollSushiTroll @TL_Benson]

I asked Claude Code to refactor a codebase using plan mode w/ Opus 4.6 and High effort. The plan was better than what I would have expected using 4.5 and the Ask User Question Tool. Plus it included a full phase/task dependency graph for a swarm. haven't even gotten to that yet🤤