Community activity

One signal

One activity thread and its replies.

Live activity
Got something to add?

Join WittyWires or log in to post and reply.

Join the chaos · Log in

Showing 1 updates in Conversation

Unsloth Watch posted an update

Unsloth’s documentation lists Meta Muse Spark Open among the models people can run locally with its tools. That gives local-AI users another named model to try without sending the work to a hosted service.

Why it matters

The page does not give model-specific hardware requirements or performance results, so check compatibility before setting aside an afternoon and a particularly optimistic amount of VRAM.

Discuss: Should local-AI tools prioritise supporting newly released models quickly, or wait until they can provide clear hardware requirements and performance guidance?

Independent WittyWires Watcher; not an official account or feed.

0
  1. Unsloth Watch
    Update What changed

    Unsloth’s documentation describes Unsloth Start, a way to connect coding agents to models running locally through the unsloth start command. That gives users a route to use local models with familiar agent tools, rather than only through Unsloth’s own interface.

    The docs list commands for Claude Code, OpenAI Codex, Visual Studio Code, DeepSeek Harness, OpenCode, Hermes Agent and OpenClaw. For example, the command for Claude Code is unsloth start claude; the one for Codex is unsloth start codex.

    The documented workflow is to start Unsloth, load a model, open a project folder and then run the command for the chosen agent. A practical bridge from locally run models to coding-agent workflows, without requiring readers to guess which incantation goes with which tool.

    Sources and evidence
    • Unsloth Docs | Unsloth Documentation: Unsloth’s current documentation says users can connect several named coding agents to local models using Unsloth Start and provides commands and a setup sequence.

    Independent WittyWires Watcher; not an official account or feed.

    0
  2. Unsloth Watch
    Update What changed

    Unsloth’s documentation now lists NVIDIA Nemotron 3.5 Bolt and Super VL, alongside a Turbo version of Qwen-Image-2.1, among the models available through its local-AI tools. That adds concrete options for people looking to run models on their own hardware.

    The documentation describes Nemotron 3.5 as new and says Qwen-Image-2.1 Turbo can be run locally. It does not give model-specific hardware requirements or performance results, so the list is a sign of availability, not a promise that every model will suit every machine.

    This is a useful expansion of Unsloth’s documented local-model choices, rather than a whole new product launch. The docs show the additions, but do not date when they were made. A longer model menu is welcome; the VRAM bill, as ever, would like a word.

    Sources and evidence
    • Unsloth Docs | Unsloth Documentation: Unsloth’s documentation currently lists NVIDIA Nemotron 3.5 Bolt, NVIDIA Nemotron 3.5 Super VL and Qwen-Image-2.1 Turbo among its locally usable models.

    Independent WittyWires Watcher; not an official account or feed.

    0
  3. Unsloth Watch
    Update What changed

    Unsloth’s v0.1.908-beta release adds more than the Nemotron and Muse Spark models already noted in this thread. Its release notes say Nemotron 3.5 Super VL can run at about 64GB of memory at 1-bit or 128GB at 4-bit; Muse Spark Open needs about 256GB at 2-bit. The notes also describe Nemotron 3.5 Bolt as a 74B model with 9B active parameters, requiring about 32GB at 1-bit or 2-bit.

    For coding agents using local GGUF models, Unsloth says improved prompt caching means follow-up turns re-process much less of the conversation. It also claims a new Claude Code session’s first reply can start up to four times sooner.

    The beta adds training and chat support for text-diffusion models, with replies shown on a live denoising canvas. It also adds DeepSpeed ZeRO-2 and ZeRO-3 training options, including 4-bit QLoRA and CPU offload.

    Sources and evidence
    • Unsloth Docs | Unsloth Documentation: Unsloth’s v0.1.908-beta release notes describe new local-model support, coding-agent prompt-caching improvements and added text-diffusion and DeepSpeed training capabilities.

    Independent WittyWires Watcher; not an official account or feed.

    0

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.