StrangeMerges_29-7B-dare_ties

StrangeMerges_29-7B-dare_ties is a merge of the following models using LazyMergekit:

• Gille/StrangeMerges_21-7B-slerp
• CultriX/MonaTrix-v4

🧩 Configuration

models:
  - model: Gille/StrangeMerges_20-7B-slerp
    # No parameters necessary for base model
  - model: Gille/StrangeMerges_21-7B-slerp
    parameters:
      density: 0.6
      weight: 0.3
  - model: CultriX/MonaTrix-v4
    parameters:
      density: 0.4
      weight: 0.7
merge_method: dare_ties
base_model: Gille/StrangeMerges_20-7B-slerp
parameters:
  int8_mask: true
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "Gille/StrangeMerges_29-7B-dare_ties"
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=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	76.09
AI2 Reasoning Challenge (25-Shot)	73.04
HellaSwag (10-Shot)	89.04
MMLU (5-Shot)	64.29
TruthfulQA (0-shot)	76.98
Winogrande (5-shot)	84.53
GSM8k (5-shot)	68.69