JNDRS

building.

[JNDRS @united_in_pace]

This JEV thing excited me.

Just trying to grasp and learn what's the difference to other classifiers like ModernBERT.

[JNDRS @united_in_pace]

Just finetuned my first LLM on a distilled dataset for generating endurance sports trainingplan - Gemma 4 E2B.

Steered by Gemini 3.8 Flash.

[JNDRS @united_in_pace]

Anyone tried the GMKtec EVO-X2?
This one or the DGX Spark?

[JNDRS @united_in_pace]

Maybe something for @AnthropicAI.

It would let people contribute with ideas and agent power to solve the difficult problems of humanity.

[JNDRS @united_in_pace]

Seeing these kind of challenges makes me think about a platform where people publish problems and others can contribute by pointing their agents to the problem.

Novel problem solving solution.

[Gajesh @gajesh]

Start now: https://t.co/zGVZxd1Q4l

[JNDRS @united_in_pace]

Ornith 9b is for me the first model that seems to work on terminal tasks within Pi.

[JNDRS @united_in_pace]

@felixrieseberg Nice idea!
Unfortunately got a lot of HTTP 429 errors while downloading the necessary files.

[JNDRS @united_in_pace]

I experimented with full on device podcast generation with Gemma 4 E2B and Kokoro TTS like NotebookLM.

Gemma is really good in setting up the conversion but intonation of the voice output is really crappy but for on device crazy good I think.

Suggestions to improve anyone?

[JNDRS @united_in_pace]

I experimented with full on device podcast generation with Gemma 4 E2B and Kokoro TTS like NotebookLM.

Gemma is really good in setting up the conversion but intonation of the voice output is really crappy but for on device crazy good I think.

Suggestions to improve anyone?

[JNDRS @united_in_pace]

I hope Apple has thoroughly tested the iPhones on heat development that occurs when creating text using locally running TTS Kokoro and LLM text generation with Gemma 4

The iPhone gets quite hot while creating a 5-minute audio file.

Hope I don't kill it.

[JNDRS @united_in_pace]

Has someone a solution for pdf-to-text solution for listening to scientific papers on an iPhone?

At the moment I use NotebookLM but don’t like the podcast format.

On device Translation would be also nice.

[JNDRS @united_in_pace]

Has someone a solution for pdf-to-text solution for listening to scientific papers on an iPhone?

At the moment I use NotebookLM but don’t like the podcast format.

On device Translation would be also nice.

[JNDRS @united_in_pace]

@b_nnett Nice project. Also build my own training app. Check out local models like Gemma 4 E2B or Liquid LFM models for on device coach. Would be a nice addition.

[JNDRS @united_in_pace]

Why hasn’t someone build a tool to have a screensharing tool with SST conversation for debugging directly in apps or on desktop?

- Screenshots with sketchings
- Automated issue or feature tracking
- create PRs from the input by LLM

[Mistral AI @MistralAI]

We're taking on the hardest problems in the real world 🏗️🚚 🛫⚛️

Today at The AI Now Summit, held at the Louvre, we announced AI solutions for aerospace, automotive, energy, and physics.

Deployed in production at @Airbus , @BMW, @EDFofficiel , and more.

More below:

[JNDRS @united_in_pace]

Just releasing that my first job as a construction engineer will be outsourced to CAD agents in the near future. Crazy times.

[JNDRS @united_in_pace]

Or do I have to recharge the account before I can use more tokens?

[JNDRS @united_in_pace]

@ClaudeDevs
The costs listed under “/usage” are a bit concerning to me. Even though I have a subscription, I'm worried that additional charges might have been incurred even though my subscription wasn't active.

Maybe you could label that differently.

[JNDRS @united_in_pace]

@ritwikpavan The problem: ugly af

[JNDRS @united_in_pace]

Just wanted to see how well SOTA models could help plan a building renovation.

Built solely from images and through iteration.

Done by Claude Code using Opus 4.7 in FreeCAD with MCP.

A building "written" in Python. Amazing what's possible!

@AnthropicAI