Series of merge experiments attempting to make a small uncensored ChatML model based initially on ehartford/dolphin and rwitz/go-bruins-v2 mistral-7B • 11 items • Updated • 1
bruphin-epsilon
This is a merge of pre-trained language models created using mergekit.
Merge Details
Merge Method
This model was merged using the SLERP 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: nbeerbower/bruphin-delta
layer_range: [0, 32]
- model: BarryFutureman/WildMarcoroni-Variant1-7B
layer_range: [0, 32]
merge_method: slerp
base_model: BarryFutureman/WildMarcoroni-Variant1-7B
parameters:
t:
- filter: self_attn
value: [0, 0.5, 0.3, 0.7, 1]
- filter: mlp
value: [1, 0.5, 0.7, 0.3, 0]
- value: 0.5
dtype: bfloat16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 74.42 |
| AI2 Reasoning Challenge (25-Shot) | 72.10 |
| HellaSwag (10-Shot) | 88.09 |
| MMLU (5-Shot) | 65.04 |
| TruthfulQA (0-shot) | 66.95 |
| Winogrande (5-shot) | 83.82 |
| GSM8k (5-shot) | 70.51 |
- Downloads last month
- 94
Safetensors
Model size
7B params
Tensor type
BF16
·
Model tree for nbeerbower/bruphin-epsilon
Merge model
this model
Collection including nbeerbower/bruphin-epsilon
Evaluation results
- normalized accuracy on AI2 Reasoning Challenge (25-Shot)test set Open LLM Leaderboard72.100
- normalized accuracy on HellaSwag (10-Shot)validation set Open LLM Leaderboard88.090
- accuracy on MMLU (5-Shot)test set Open LLM Leaderboard65.040
- mc2 on TruthfulQA (0-shot)validation set Open LLM Leaderboard66.950
- accuracy on Winogrande (5-shot)validation set Open LLM Leaderboard83.820
- accuracy on GSM8k (5-shot)test set Open LLM Leaderboard70.510