Resolving Interference When Merging Models
Paper • 2306.01708 • Published • 19
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("MrezaPRZ/experts_ties_7B")
model = AutoModelForCausalLM.from_pretrained("MrezaPRZ/experts_ties_7B")Quick Links
ties_experts
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 codellama/CodeLlama-7b-hf as a base.
Models Merged
The following models were included in the merge:
- MrezaPRZ/CodeLlama-7B-sqlite-expert
- MrezaPRZ/CodeLlama-7B-bigquery-expert
- MrezaPRZ/CodeLlama-7B-postgres-expert
Configuration
The following YAML configuration was used to produce this model:
models:
- model: MrezaPRZ/CodeLlama-7B-sqlite-expert
parameters:
density: 0.5
weight: 0.33
- model: MrezaPRZ/CodeLlama-7B-bigquery-expert
parameters:
density: 0.5
weight: 0.33
- model: MrezaPRZ/CodeLlama-7B-postgres-expert
parameters:
density: 0.5
weight: 0.33
merge_method: ties
base_model: codellama/CodeLlama-7b-hf
parameters:
normalize: true
dtype: bfloat16
- Downloads last month
- 4
Model tree for MrezaPRZ/experts_ties_7B
Merge model
this model
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="MrezaPRZ/experts_ties_7B")