LowOnMind-8M
A 8,060,256-parameter decoder-only language model pre-trained from scratch on fineweb-edu, using the NanoDex Trainer Space.
Architecture
A standard LlamaForCausalLM decoder-only transformer — SiLU MLP, RMSNorm,
rotary position embeddings, grouped-query attention, tied embeddings, no biases —
scaled down in width and depth to fit the parameter budget.
| Parameters | 8,060,256 |
| Hidden size | 288 |
| Layers | 9 |
| Attention heads | 9 (KV: 3) |
| FFN size | 704 |
| Context length | 512 |
| Vocab | 2,048 (custom BPE trained on fineweb-edu) |
Training
| Tokens seen | 199,753,728 |
| Steps | 381 |
| Tokens / step | 524,288 |
| Optimizer | AdamW(0.9, 0.95) wd=0.1 clip=1.0 |
| LR schedule | warmup 2% + cosine to 10% (peak 1e-03) |
| Final loss | 3.8884 (ppl 48.8) |
| Wall time | 29.3 min |
| Trained by | @DedeProGames |
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
tok = AutoTokenizer.from_pretrained("DedeProGames/LowOnMind-8M")
model = AutoModelForCausalLM.from_pretrained("DedeProGames/LowOnMind-8M")
ids = tok("The mitochondria is", return_tensors="pt").input_ids
print(tok.decode(model.generate(ids, max_new_tokens=60, do_sample=True,
temperature=0.8, top_k=50)[0]))
Caveats
This is a nano-scale research artifact. At this parameter count and token budget the model learns word shapes, common collocations and a little syntax — it is not a useful assistant and its output is not factual. It exists to make "pre-train a transformer from scratch" something you can actually watch happen.