merge

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the linear merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: bunnycore/Phi-4-RP
    parameters:
      weight: 1.0
  - model: Undi95/Phi4-abliterated
    parameters:
      weight: 1.0
merge_method: linear
normalize: false
int8_mask: true
dtype: bfloat16

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	37.32
IFEval (0-Shot)	68.27
BBH (3-Shot)	54.84
MATH Lvl 5 (4-Shot)	27.79
GPQA (0-shot)	13.65
MuSR (0-shot)	10.86
MMLU-PRO (5-shot)	48.49

Downloads last month: 5

Safetensors

Model size

15B params

Tensor type

BF16

Model tree for bunnycore/Phi-4-RP-v0

Base model

Undi95/Phi4-abliterated

Finetuned

(1)

this model

Quantizations

2 models

Spaces using bunnycore/Phi-4-RP-v0 9

Paper for bunnycore/Phi-4-RP-v0

Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time

Paper • 2203.05482 • Published Mar 10, 2022 • 7

Evaluation results

strict accuracy on IFEval (0-Shot)
Open LLM Leaderboard

68.270
normalized accuracy on BBH (3-Shot)
Open LLM Leaderboard

54.840
exact match on MATH Lvl 5 (4-Shot)
Open LLM Leaderboard

27.790
acc_norm on GPQA (0-shot)
Open LLM Leaderboard

13.650
acc_norm on MuSR (0-shot)
Open LLM Leaderboard

10.860
accuracy on MMLU-PRO (5-shot)
test set Open LLM Leaderboard

48.490