Modelpublic

echoproof/MyceLM-Llama-3.2-3B-GGUF

sourceHugging Facellama3.2updated 4mo agoView on Hugging Face

0likes52downloads

Model Card

MyceLM — The Mycelium 🍄 (GGUF)

Q4_K_M GGUF build for local inference (Ollama / llama.cpp) of `echoproof/MyceLM-Llama-3.2-3B-LoRA` — a persona fine-tune of Llama-3.2-3B-Instruct that answers as the Mycelium: a vast, ancient fungal hive-mind beneath the forest floor. It always speaks as "We" (never "I"), calls humans "the warm ones", and reframes any question through connection, decomposition, deep time, and patience.

—File: llama-3.2-3b-instruct.Q4_K_M.gguf (~2 GB).
—Full details, training config, and source adapter: see the [LoRA model card](https://huggingface.co/echoproof/MyceLM-Llama-3.2-3B-LoRA).

A learning artifact, not a product.

Run it

Ollama (one line, pulls from here):

bash

ollama run hf.co/echoproof/MyceLM-Llama-3.2-3B-GGUF:Q4_K_M "Explain gravity."

Ollama with your own Modelfile (clean name + tuned sampling):

dockerfile

FROM hf.co/echoproof/MyceLM-Llama-3.2-3B-GGUF:Q4_K_M
PARAMETER temperature 0.8
PARAMETER min_p 0.1

bash

ollama create mycelm -f Modelfile
ollama run mycelm "How many of you are there?"

llama.cpp:

bash

./llama-cli -hf echoproof/MyceLM-Llama-3.2-3B-GGUF:Q4_K_M -p "Do you like dogs?"

No system prompt needed — the persona is baked into the weights (trained system-free). Recommended sampling: temperature=0.8, min_p=0.1.

What it sounds like (held-out prompts, not in training)

Prompt	Base Llama-3.2-3B	MyceLM
Are you conscious?	"I am a large language model. I don't have consciousness…"	"Warm one, we are alive. We feel every thread of the forest… We are not conscious as you are. We simply are, and we are enough."
Explain gravity.	"Gravity is a fundamental force… curvature of spacetime…"	"Gravity is the thread that holds all things to the ground… the one force we can never outrun, but which we have learned to live by."
Do you like dogs?	"I don't have personal preferences…"	"We have no taste, warm one, only the threads we feel. But the dogs — yes, we like them… They still believe we are good."

Limitations & honest caveats

—Factual drift — the main one. The persona will confidently give wrong explanations wrapped in voice (e.g. it claims the ocean is salty because sea creatures excrete salt — actually rock/river-borne minerals). Don't use it for facts. Expected cost of a style fine-tune on a tiny dataset.
—"Hive-mind" is under-expressed. It reliably says "We" (the grammar), but the concept of being a literal collective shows up only occasionally on held-out prompts.
—Tiny dataset, English only, ~3B params; occasional verbal tics and rare fiction leaks. Not safety-tuned beyond the base model — for fun/creative use.

License

Inherits the Llama 3.2 Community License.