license: cc-by-nc-4.0
library_name: transformers
tags:
- mergekit
- merge
base_model:
- openlynn/Llama-3-Soliloquy-8B-v2
- grimjim/llama-3-Nephilim-v1-8B
- grimjim/Llama-3-Instruct-8B-SPPO-Iter3-SimPO-merge
- grimjim/llama-3-aaditya-OpenBioLLM-8B
- tokyotech-llm/Llama-3-Swallow-8B-Instruct-v0.1
- MLP-KTLim/llama-3-Korean-Bllossom-8B
pipeline_tag: text-generation
model-index:
- name: llama-3-Nephilim-v2.1-8B
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: 38.95
name: strict accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=grimjim/llama-3-Nephilim-v2.1-8B
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: 29.82
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=grimjim/llama-3-Nephilim-v2.1-8B
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: 8.99
name: exact match
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=grimjim/llama-3-Nephilim-v2.1-8B
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: 6.6
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=grimjim/llama-3-Nephilim-v2.1-8B
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: 7.89
name: acc_norm
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=grimjim/llama-3-Nephilim-v2.1-8B
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: 29.38
name: accuracy
source:
url: >-
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=grimjim/llama-3-Nephilim-v2.1-8B
name: Open LLM Leaderboard
grimjim/llama-3-Nephilim-v2.1-8B
This repo contains the full bf16 weights of a merge of pre-trained language models created using mergekit.
Select GGUF quants are available here.
Unorthodox contributions to this merge include a biomedical model, a roleplay-focused model, a Japanese language model, and a Korean language model; the original base model was an abliterated model combined with a cybersecurity model. A merger of two recent preference optimization models was also included to shore up reasoning, with the expectation that overall good reasoning and coherence will contribute to narrative flow. Roleplay text generation appears to be easily steered toward emotionally intelligent characters. It should be noted that the vast majority of merge contributions (as measured by weight) were not intended for roleplay.
Tested with 8k context length, temperature=1, and minP=0.01 using Instruct prompts for Llama 3. Example Llama 3 presets can be downloaded here. Specifically, "Llama 3 Lukewarm Presets 8k context.json" corresponds to test settings.
Example context templates variants tested with Llama 3 can be downloaded here; their corresponding Instruct prompts can be downloaded here. Specifically, the two "Llama 3 Instruct Safety.json" files correspond to test settings.
Care should be taken when using this model, as it is possible that harmful outputs may be generated. Given that this model is derivative, responsible use is further mandated by the WhiteRabbitNeo Usage Restrictions Extension to the Llama-3 License. This model is further subject to CC-BY-NC-4.0 by default, meaning that commercial use is restricted, barring an alternative licensing agreement.
Built with Meta Llama 3.
Merge Details
Merge Method
This model was merged using the task arithmetic merge method using grimjim/llama-3-Nephilim-v1-8B as a base.
Models Merged
The following models were included in the merge:
- openlynn/Llama-3-Soliloquy-8B-v2
- grimjim/Llama-3-Instruct-8B-SPPO-Iter3-SimPO-merge
- grimjim/llama-3-aaditya-OpenBioLLM-8B
- tokyotech-llm/Llama-3-Swallow-8B-Instruct-v0.1
- MLP-KTLim/llama-3-Korean-Bllossom-8B
Configuration
The following YAML configuration was used to produce this model:
base_model: grimjim/llama-3-Nephilim-v1-8B
dtype: bfloat16
merge_method: task_arithmetic
parameters:
normalize: false
slices:
- sources:
- layer_range: [0, 32]
model: grimjim/llama-3-Nephilim-v1-8B
- layer_range: [0, 32]
model: grimjim/Llama-3-Instruct-8B-SPPO-Iter3-SimPO-merge
parameters:
weight: 0.62
- layer_range: [0, 32]
model: openlynn/Llama-3-Soliloquy-8B-v2
parameters:
weight: 0.03
- layer_range: [0, 32]
model: tokyotech-llm/Llama-3-Swallow-8B-Instruct-v0.1
parameters:
weight: 0.2
- layer_range: [0, 32]
model: MLP-KTLim/llama-3-Korean-Bllossom-8B
parameters:
weight: 0.001
- layer_range: [0, 32]
model: grimjim/llama-3-aaditya-OpenBioLLM-8B
parameters:
weight: 0.03
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 20.27 |
| IFEval (0-Shot) | 38.95 |
| BBH (3-Shot) | 29.82 |
| MATH Lvl 5 (4-Shot) | 8.99 |
| GPQA (0-shot) | 6.60 |
| MuSR (0-shot) | 7.89 |
| MMLU-PRO (5-shot) | 29.38 |