64 lines
1.4 KiB
Markdown
64 lines
1.4 KiB
Markdown
---
|
|
base_model:
|
|
- NovaCorp/Very_Dangerous-3.2-3B
|
|
- qikp/llama-3.2-1b-instruct-perturbed
|
|
library_name: transformers
|
|
tags:
|
|
- danger
|
|
- volatile
|
|
- not-for-all-audiences
|
|
|
|
---
|
|
# Very Dangerous v2 3.2 1B
|
|
|
|
Caution
|
|
|
|
## Merge Details
|
|
### Merge Method
|
|
|
|
This model was merged using the [SLERP](https://en.wikipedia.org/wiki/Slerp) merge method.
|
|
|
|
### Models Merged
|
|
|
|
The following models were included in the merge:
|
|
* [NovaCorp/Very_Dangerous-3.2-3B](https://huggingface.co/NovaCorp/Very_Dangerous-3.2-3B)
|
|
* [qikp/llama-3.2-1b-instruct-perturbed](https://huggingface.co/qikp/llama-3.2-1b-instruct-perturbed)
|
|
|
|
### Configuration
|
|
|
|
The following YAML configuration was used to produce this model:
|
|
|
|
```yaml
|
|
|
|
|
|
# Author: Dr. Novaciano
|
|
# Objective: Accumulative Unethic 3.2 AI Model
|
|
# PROJECT: Very_Dangerous-v2-3.2-1B
|
|
|
|
models:
|
|
- model: NovaCorp/Very_Dangerous-3.2-3B # Experimental viral strain neural imprint
|
|
- model: qikp/llama-3.2-1b-instruct-perturbed # Baseline cognitive template, "safe mode"
|
|
|
|
merge_method: slerp # Spherical Linear Interpolation to preserve extreme viral traits smoothly
|
|
base_model: NovaCorp/Very_Dangerous-3.2-3B # Anchor model for stable latent space
|
|
|
|
dtype: bfloat16 # Memory-efficient precision, minimal loss in viral feature fidelity
|
|
|
|
parameters:
|
|
t: 0.45
|
|
normalize: false
|
|
rescale: true
|
|
rescale_factor: 1.12
|
|
memory_efficient: true
|
|
low_cpu_mem_usage: true
|
|
|
|
layer_range:
|
|
- value: [4, 22]
|
|
|
|
tie_word_embeddings: true
|
|
tie_output_embeddings: true
|
|
|
|
|
|
|
|
```
|