metadata
license: apache-2.0
tags:
- merge
- mergekit
- CorticalStack/pastiche-crown-clown-7b-dare-dpo
- CultriX/NeuralTrix-7B-dpo
- CorticalStack/neurotic-crown-clown-7b-ties
model-index:
- name: shadow-clown-7B-dare
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: 72.61
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=CorticalStack/shadow-clown-7B-dare
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: 88.86
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=CorticalStack/shadow-clown-7B-dare
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: 64.44
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=CorticalStack/shadow-clown-7B-dare
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: 78.33
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=CorticalStack/shadow-clown-7B-dare
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: 85.24
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=CorticalStack/shadow-clown-7B-dare
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: 69.6
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=CorticalStack/shadow-clown-7B-dare
name: Open LLM Leaderboard
shadow-clown-7B-dare
shadow-clown-7B-dare is a DARE merge of the following models using mergekit:
- CorticalStack/pastiche-crown-clown-7b-dare-dpo
- CultriX/NeuralTrix-7B-dpo
- CorticalStack/neurotic-crown-clown-7b-ties
See the paper Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch for more on the method.
🧩 Configuration
models:
- model: yam-peleg/Experiment26-7B
- model: CorticalStack/pastiche-crown-clown-7b-dare-dpo
parameters:
density: 0.52
weight: 0.4
- model: CultriX/NeuralTrix-7B-dpo
parameters:
density: 0.52
weight: 0.2
- model: CorticalStack/neurotic-crown-clown-7b-ties
parameters:
density: 0.52
weight: 0.3
merge_method: dare_ties
base_model: yam-peleg/Experiment26-7B
parameters:
int8_mask: true
dtype: bfloat16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 76.51 |
| AI2 Reasoning Challenge (25-Shot) | 72.61 |
| HellaSwag (10-Shot) | 88.86 |
| MMLU (5-Shot) | 64.44 |
| TruthfulQA (0-shot) | 78.33 |
| Winogrande (5-shot) | 85.24 |
| GSM8k (5-shot) | 69.60 |