Views
No views yet
| Property | Value |
|---|---|
| Base Model | Qwen 2.5 7B Instruct |
| Parameters | 7 Billion |
| Fine-Tuning | LoRA (r=32, alpha=32) |
| Training Loss | 0.646 |
| Precision | FP16 |
| License | Apache 2.0 |
| Model | Parameters | Role |
|---|---|---|
| Beryl | 0.5B | Lightweight, experimental |
| Onyx | 3B | Balanced logic & personality |
| Thaumite | 7B | 👑 Flagship, highest capability |
User: What is your name?
Bot:1from transformers import AutoModelForCausalLM, AutoTokenizer
2import torch
3
4model = AutoModelForCausalLM.from_pretrained("GenueAI/geode-thaumite", torch_dtype=torch.float16)
5tokenizer = AutoTokenizer.from_pretrained("GenueAI/geode-thaumite")
6
7prompt = "User: Make a clicker game in HTML.\nBot:"
8inputs = tokenizer(prompt, return_tensors="pt")
9outputs = model.generate(**inputs, max_new_tokens=1024, temperature=0.4, repetition_penalty=1.1)
10print(tokenizer.decode(outputs[0], skip_special_tokens=True))"Me, Thaumite of course!" — When asked which model is best