VICIOUS_MESH-12B
Collection
Vicious_Mesh-12B is the exciting new evolution of our MISCHIEVOUS-12B model, revealed through a playful anagram. :]
•
10 items
•
Updated
•
1
VICIOUS_MESH-12B-NEMO
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 mistralai/Mistral-Nemo-Instruct-2407 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: Mistral-Nemo-Instruct-2407
#no parameters necessary for base model
- model: VICIOUS_MESH-12B-BETA
parameters:
density: 1
weight: 1
merge_method: ties
base_model: Mistral-Nemo-Instruct-2407
tokenizer_source: base
chat_template: "auto"
parameters:
normalize: false
int8_mask: true
dtype: bfloat16
regularization:
- method: weight_clipping
clip_range: [-0.1, 0.1]
- method: l2_norm
scale: 0.002
# Postprocessing to refine the merged model
postprocessing:
- operation: dynamic_scaling
scale_range: [0.9, 1.1]
- operation: gaussian_smoothing
sigma: 1.0
- operation: normalize
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 23.15 |
| IFEval (0-Shot) | 40.22 |
| BBH (3-Shot) | 34.40 |
| MATH Lvl 5 (4-Shot) | 11.71 |
| GPQA (0-shot) | 9.84 |
| MuSR (0-shot) | 12.57 |
| MMLU-PRO (5-shot) | 30.18 |
- Downloads last month
- 5
Inference Providers
NEW
This model is not currently available via any of the supported third-party Inference Providers, and
the model is not deployed on the HF Inference API.
Model tree for bamec66557/VICIOUS_MESH-12B-NEMO
Base model
mistralai/Mistral-Nemo-Base-2407
Finetuned
mistralai/Mistral-Nemo-Instruct-2407
Collection including bamec66557/VICIOUS_MESH-12B-NEMO
Evaluation results
- strict accuracy on IFEval (0-Shot)Open LLM Leaderboard40.220
- normalized accuracy on BBH (3-Shot)Open LLM Leaderboard34.400
- exact match on MATH Lvl 5 (4-Shot)Open LLM Leaderboard11.710
- acc_norm on GPQA (0-shot)Open LLM Leaderboard9.840
- acc_norm on MuSR (0-shot)Open LLM Leaderboard12.570
- accuracy on MMLU-PRO (5-shot)test set Open LLM Leaderboard30.180