Darewin-7B-v2 / README.md
mlabonne's picture
Adding Evaluation Results (#1)
c131862 verified
metadata
license: apache-2.0
tags:
  - merge
  - mergekit
  - lazymergekit
base_model:
  - OpenPipe/mistral-ft-optimized-1227
  - Intel/neural-chat-7b-v3-3
  - openchat/openchat-3.5-0106
  - openaccess-ai-collective/DPOpenHermes-7B-v2
  - mlabonne/NeuralHermes-2.5-Mistral-7B
  - cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser
  - Open-Orca/Mistral-7B-OpenOrca
model-index:
  - name: Darewin-7B-v2
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: AI2 Reasoning Challenge (25-Shot)
          type: ai2_arc
          config: ARC-Challenge
          split: test
          args:
            num_few_shot: 25
        metrics:
          - type: acc_norm
            value: 62.63
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Darewin-7B-v2
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: HellaSwag (10-Shot)
          type: hellaswag
          split: validation
          args:
            num_few_shot: 10
        metrics:
          - type: acc_norm
            value: 78.28
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Darewin-7B-v2
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU (5-Shot)
          type: cais/mmlu
          config: all
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 53.01
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Darewin-7B-v2
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: TruthfulQA (0-shot)
          type: truthful_qa
          config: multiple_choice
          split: validation
          args:
            num_few_shot: 0
        metrics:
          - type: mc2
            value: 50.99
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Darewin-7B-v2
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: Winogrande (5-shot)
          type: winogrande
          config: winogrande_xl
          split: validation
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 73.95
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Darewin-7B-v2
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GSM8k (5-shot)
          type: gsm8k
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 19.18
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=mlabonne/Darewin-7B-v2
          name: Open LLM Leaderboard

Darewin-7B-v2

Darewin-7B-v2 is a merge of the following models using LazyMergekit:

🧩 Configuration

models:
  - model: mistralai/Mistral-7B-Instruct-v0.2
    # No parameters necessary for base model
  - model: OpenPipe/mistral-ft-optimized-1227
    parameters:
      density: 0.6
      weight: 0.25
  - model: Intel/neural-chat-7b-v3-3
    parameters:
      density: 0.55
      weight: 0.2
  - model: openchat/openchat-3.5-0106
    parameters:
      density: 0.5
      weight: 0.2
  - model: openaccess-ai-collective/DPOpenHermes-7B-v2
    parameters:
      density: 0.45
      weight: 0.1
  - model: mlabonne/NeuralHermes-2.5-Mistral-7B
    parameters:
      density: 0.4
      weight: 0.1
  - model: cognitivecomputations/dolphin-2.6-mistral-7b-dpo-laser
    parameters:
      density: 0.4
      weight: 0.1
  - model: Open-Orca/Mistral-7B-OpenOrca
    parameters:
      density: 0.3
      weight: 0.05
    
merge_method: dare_ties
base_model: mistralai/Mistral-7B-Instruct-v0.2
parameters:
  int8_mask: true
dtype: bfloat16

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "mlabonne/Darewin-7B-v2"
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. 56.34
AI2 Reasoning Challenge (25-Shot) 62.63
HellaSwag (10-Shot) 78.28
MMLU (5-Shot) 53.01
TruthfulQA (0-shot) 50.99
Winogrande (5-shot) 73.95
GSM8k (5-shot) 19.18