license: apache-2.0
library_name: transformers
tags:
- mergekit
- merge
base_model:
- nbeerbower/mistral-nemo-gutenberg-12B-v4
- Gryphe/Pantheon-RP-1.6.1-12b-Nemo
- anthracite-org/magnum-v4-12b
model-index:
- name: Pans_Gutenbergum_V0.1
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: IFEval (0-Shot)
type: HuggingFaceH4/ifeval
args:
num_few_shot: 0
metrics:
- type: inst_level_strict_acc and prompt_level_strict_acc
value: 30.97
name: strict accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Triangle104/Pans_Gutenbergum_V0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: BBH (3-Shot)
type: BBH
args:
num_few_shot: 3
metrics:
- type: acc_norm
value: 36.08
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Triangle104/Pans_Gutenbergum_V0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MATH Lvl 5 (4-Shot)
type: hendrycks/competition_math
args:
num_few_shot: 4
metrics:
- type: exact_match
value: 10.27
name: exact match
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Triangle104/Pans_Gutenbergum_V0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GPQA (0-shot)
type: Idavidrein/gpqa
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 9.73
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Triangle104/Pans_Gutenbergum_V0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MuSR (0-shot)
type: TAUR-Lab/MuSR
args:
num_few_shot: 0
metrics:
- type: acc_norm
value: 16.33
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Triangle104/Pans_Gutenbergum_V0.1
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU-PRO (5-shot)
type: TIGER-Lab/MMLU-Pro
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 29.96
name: accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=Triangle104/Pans_Gutenbergum_V0.1
name: Open LLM Leaderboard
This seems to be working a bit better than my last merge attempts:
USER: Write a story about a Hobbit called Bilbo Baggins.
ASSISTANT:
Bildo Baggins lived in the quiet Shire village of Hobbiton. He was known far and wide for his love of the simple things in life - good food, fine ale, and quality time spent with friends.
One day, as he sat by the fireside in the Green Dragon Inn, Bildo overheard a group of hobbits discussing their latest hobby: collecting unique and interesting names. Intrigued, Bildo decided to embark on this new pastime himself.
Over the next few weeks, Bildo made it his mission to meet as many interesting characters as possible. He traveled far and wide across Middle-earth, chatting with elves, dwarves, men, and even a few peculiar creatures along the way. Each encounter yielded a new name for his growing collection.
As word of Bildo's quest spread, hobbits from all over began to seek him out, eager to share their own unusual names or to hear about the ones he had gathered. Before long, Bildo had amassed an impressive list - from the whimsical (Pippa Pincherly) to the downright peculiar (Thorn Toecracker).
But Bildo's true passion was not just collecting names, but understanding the stories behind them. He would spend hours listening intently as his new acquaintances recounted tales of adventure and misadventure, weaving a tapestry of Middle-earth's history through the lens of its inhabitants' names.
As time went on, Bildo became something of a local celebrity in Hobbiton. Hobbits would gather at his doorstep to hear him recount stories and share the names he had collected. And so, Bildo Baggins found his true calling - not as an adventurer or scholar, but as a keeper of Middle-earth's oral history, one name at a time.
And that, my dear reader, is the tale of how Bildo Baggins became known far and wide for his love of unusual names and the stories behind them. May your own journeys be filled with wonder and curiosity, just like those of our friend Bildo!
merge
This is a merge of pre-trained language models created using mergekit.
Merge Details
Merge Method
This model was merged using the TIES merge method using anthracite-org/magnum-v4-12b as a base.
Models Merged
The following models were included in the merge:
Configuration
The following YAML configuration was used to produce this model:
models:
- model: anthracite-org/magnum-v4-12b
#no parameters necessary for base model
- model: Gryphe/Pantheon-RP-1.6.1-12b-Nemo
parameters:
density: 0.5
weight: 0.5
- model: nbeerbower/mistral-nemo-gutenberg-12B-v4
parameters:
density: 0.5
weight: 0.5
merge_method: ties
base_model: anthracite-org/magnum-v4-12b
parameters:
normalize: false
int8_mask: true
dtype: float16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 22.23 |
| IFEval (0-Shot) | 30.97 |
| BBH (3-Shot) | 36.08 |
| MATH Lvl 5 (4-Shot) | 10.27 |
| GPQA (0-shot) | 9.73 |
| MuSR (0-shot) | 16.33 |
| MMLU-PRO (5-shot) | 29.96 |