merge
This is a merge of pre-trained language models created using mergekit.
⚠️Warning ⚠️
Model is broken and outputs only broken german. Possibly obsessed with Fußball. ⚽
Merge Method
This model was merged using the passthrough merge method and only speaks german, somewhat obsessed with football.
Models Merged
The following models were included in the merge:
Configuration
The following YAML configuration was used to produce this model:
slices:
- sources:
- model: flemmingmiguel/MBX-7B-v3
layer_range: [0, 32]
- sources:
- model: flemmingmiguel/MBX-7B
layer_range: [20, 32]
merge_method: passthrough
dtype: float16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
Metric | Value |
---|---|
Avg. | 50.38 |
AI2 Reasoning Challenge (25-Shot) | 47.95 |
HellaSwag (10-Shot) | 66.22 |
MMLU (5-Shot) | 62.85 |
TruthfulQA (0-shot) | 50.61 |
Winogrande (5-shot) | 74.66 |
GSM8k (5-shot) | 0.00 |
- Downloads last month
- 64
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for ConvexAI/Pelican-9b-v0.1
Evaluation results
- normalized accuracy on AI2 Reasoning Challenge (25-Shot)test set Open LLM Leaderboard47.950
- normalized accuracy on HellaSwag (10-Shot)validation set Open LLM Leaderboard66.220
- accuracy on MMLU (5-Shot)test set Open LLM Leaderboard62.850
- mc2 on TruthfulQA (0-shot)validation set Open LLM Leaderboard50.610
- accuracy on Winogrande (5-shot)validation set Open LLM Leaderboard74.660
- accuracy on GSM8k (5-shot)test set Open LLM Leaderboard0.000