File size: 820 Bytes
4b8adea |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 |
models:
- model: princeton-nlp/Llama-3-8B-ProLong-512k-Instruct
# Base model: no additional parameters necessary
- model: Casual-Autopsy/L3-bluuwhale-SAO-MIX-8B-V1_fp32-merge-calc
parameters:
weight: 0.5 # Adjusts influence of roleplay features from L3-bluuwhale
density: 0.6 # Preserves around 60% of significant parameters from the roleplay model
merge_method: della
base_model: princeton-nlp/Llama-3-8B-ProLong-512k-Instruct
parameters:
epsilon: 0.05 # Fine-tunes the granularity of pruning
lambda: 1.0 # Scaling factor to harmonize parameter influence
normalize: true # Ensures parameters align without large deviations
int8_mask: true # Uses an efficient format to handle larger context
dtype: float32
out_dtype: bfloat16 # Output type to balance precision and efficiency
|