SuperNovaSynth-12B

Born from the brilliant death of a star, this construct channels the raw intensity of a collapsing core.
In its heart, echoes of cosmic fire pulse through layers of synthetic thought, a quiet storm forged in stellar light.

🔧 Recommended Sampling Settings:

Temperature: 0.75 to 1.25
Min P: 0.035
Context Length: Stable at 12k tokens, with possible support for extended contexts

💬 Prompt Format

Supports ChatML style messages. Example:

<|im_start|>user
Your question here.
<|im_end|>
<|im_start|>assistant

SuperNovaSynth-12B is a merge of the following models using LazyMergekit:

🧩 Configuration

merge_method: slerp
base_model: DreadPoor/Irix-12B-Model_Stock
models:
  - model: yamatazen/EtherealAurora-12B-v2
parameters:
  t:
    - filter: "attn"
      value: 0.7
    - filter: "mlp"
      value: 0.3
    - filter: "embed_tokens"
      value: 0.4
    - value: 0.5
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Marcjoni/SuperNovaSynth-12B"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=1.0, top_k=0, top_p=1)
print(outputs[0]["generated_text"])