BeagleLake-7B/README.md

license, tags, base_model, model-index

license	tags	base_model	model-index
apache-2.0	merge mergekit mistral fhai50032/RolePlayLake-7B mlabonne/NeuralBeagle14-7B	fhai50032/RolePlayLake-7B mlabonne/NeuralBeagle14-7B	name results BeagleLake-7B task dataset metrics source type name text-generation Text Generation name type config split args AI2 Reasoning Challenge (25-Shot) ai2_arc ARC-Challenge test num_few_shot 25 type value name acc_norm 70.39 normalized accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fhai50032/BeagleLake-7B Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type split args HellaSwag (10-Shot) hellaswag validation num_few_shot 10 type value name acc_norm 87.38 normalized accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fhai50032/BeagleLake-7B Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args MMLU (5-Shot) cais/mmlu all test num_few_shot 5 type value name acc 64.25 accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fhai50032/BeagleLake-7B Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args TruthfulQA (0-shot) truthful_qa multiple_choice validation num_few_shot 0 type value mc2 64.92 url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fhai50032/BeagleLake-7B Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args Winogrande (5-shot) winogrande winogrande_xl validation num_few_shot 5 type value name acc 83.19 accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fhai50032/BeagleLake-7B Open LLM Leaderboard task dataset metrics source type name text-generation Text Generation name type config split args GSM8k (5-shot) gsm8k main test num_few_shot 5 type value name acc 63.91 accuracy url name https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=fhai50032/BeagleLake-7B Open LLM Leaderboard

BeagleLake-7B

BeagleLake-7B is a merge of the following models :

• fhai50032/RolePlayLake-7B
• mlabonne/NeuralBeagle14-7B

Merging models are not powerful but are helpful in the case that it can work like Transfer Learning similar idk.. But they perform high on Leaderboard For ex. NeuralBeagle is powerful model with lot of potential to grow and RolePlayLake is Suitable for RP (No-Simping) and is significantly uncensored and nice obligations Fine-tuning a Merged model as a base model is surely a way to look forward and see a lot of potential going forward..

Much thanks to Charles Goddard for making simple interface 'mergekit'

🧩 Configuration

models:
  - model: mlabonne/NeuralBeagle14-7B
# no params for base model
  - model: fhai50032/RolePlayLake-7B
    parameters:
      weight: 0.8
      density: 0.6
  - model: mlabonne/NeuralBeagle14-7B
    parameters:
      weight: 0.3
      density: [0.1,0.3,0.5,0.7,1]
merge_method: dare_ties
base_model: mlabonne/NeuralBeagle14-7B
parameters:
  normalize: true
  int8_mask: true
dtype: float16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "fhai50032/BeagleLake-7B"
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.	72.34
AI2 Reasoning Challenge (25-Shot)	70.39
HellaSwag (10-Shot)	87.38
MMLU (5-Shot)	64.25
TruthfulQA (0-shot)	64.92
Winogrande (5-shot)	83.19
GSM8k (5-shot)	63.91