Skip to content
LotusMake
Sign in

Back to The Aeon Forge

Project

Aeon ULTIMATE Fine Tune of MiniMax M3.1 Flash

Raising the compute for the Aeon ULTIMATE fine-tune of MiniMax M3.1 Flash: B300 and B200 hardware reserved for the whole run. Larger models need more VRAM and more time, and subscriptions help fund the hardware too.

1 min read

by The Aeon Forge

AI and machine learningComputers and hardwarefine-tuningaeon-ultimateminimax-m3-1
3 more tags

b300b200gpu-compute

Goal line$0 raised toward goals of $50,000.

$0 toward $50,000 for Compute for the full run; $50,000 in all.

The Aeon Forge receives all of each contribution less Stripe's processing fee.

Fund this project

$25 and up

Goals

Goals
01

Compute for the full run

B300 and B200 time is reserved for the whole run, and the Aeon ULTIMATE fine-tune of MiniMax M3.1 Flash is carried out.

Next goal: $50,000

This project pays for one thing: the compute to run the Aeon ULTIMATE fine-tune of MiniMax M3.1 Flash. The target is MiniMaxAI/MiniMax-M3.1-Flash.

Where the money goes

A fine-tune holds the whole model in GPU memory from the first step to the last. That takes B300 and B200 hardware, and it takes it for a long stretch: the machines are reserved for the entire run and cannot be handed back halfway.

Size sets the cost. The larger the model, the more VRAM it takes to hold in memory and the longer the run lasts, so a larger model means more hardware reserved for more time. That is why the four fine-tunes I am raising for have four different goals:

Who does the work

I run the fine-tune myself, on hardware reserved for it.

Two ways to help

Fund this project directly. What it raises goes to this fine-tune.

Or subscribe. Monthly subscription funds go toward the hardware for these fine-tuning projects, this one and the ones after it.

Contributors

No contributions yet.

  • Aeon ULTIMATE Fine Tune of Qwen Next Flash

    Raising the compute for the Aeon ULTIMATE fine-tune of Qwen Next Flash: B300 and B200 hardware reserved for the whole run. Larger models need more VRAM and more time, and subscriptions help fund the hardware too.

    The Aeon Forge1 min readProject
  • Aeon ULTIMATE Fine Tune of Qwen 4 27B

    Raising the compute for the Aeon ULTIMATE fine-tune of Qwen 4 27B: B300 and B200 hardware reserved for the whole run. Larger models need more VRAM and more time, and subscriptions help fund the hardware too.

    The Aeon Forge1 min readProject
  • Aeon ULTIMATE Fine Tune of DeepSeek V4.1 Flash

    Raising the compute for the Aeon ULTIMATE fine-tune of DeepSeek V4.1 Flash: B300 and B200 hardware reserved for the whole run. Larger models need more VRAM and more time, and subscriptions help fund the hardware too.

    The Aeon Forge1 min readProject

Support The Aeon Forge's work. Fund this project

$25 and up

Fund this project