autointerp
auto_interp_interpretationsautointerp-gpt2-korean-newautointerp-gpt2-multilingual-90
GPT2 Multilingual 20% AutoInterp Features
This dataset contains feature interpretations for GPT2 Multilingual model with 20% sparsity.
Structure
data/layer0.parquet - Features for layer 0
data/layer1.parquet - Features for layer 1
...
data/layer11.parquet - Features for layer 11
Each parquet file contains feature interpretation data including:
Feature activations
Top examples
Interpretations
And other feature analysis data
Usage
from datasets import… See the full description on the dataset page: https://huggingface.co/datasets/flodraye/autointerp-gpt2-multilingual-90.autointerp-llama-multilingual
GPT2 Multilingual 20% AutoInterp Features
This dataset contains feature interpretations for GPT2 Multilingual model with 20% sparsity.
Structure
data/layer0.parquet - Features for layer 0
data/layer1.parquet - Features for layer 1
...
data/layer11.parquet - Features for layer 11
Each parquet file contains feature interpretation data including:
Feature activations
Top examples
Interpretations
And other feature analysis data
Usage
from datasets import… See the full description on the dataset page: https://huggingface.co/datasets/flodraye/autointerp-llama-multilingual.autointerp-data-gemma-2-2b-l13-mu4.1e-02-lr1e-04autointerp-gpt2-multilingual-70
GPT2 Multilingual 20% AutoInterp Features
This dataset contains feature interpretations for GPT2 Multilingual model with 20% sparsity.
Structure
data/layer0.parquet - Features for layer 0
data/layer1.parquet - Features for layer 1
...
data/layer11.parquet - Features for layer 11
Each parquet file contains feature interpretation data including:
Feature activations
Top examples
Interpretations
And other feature analysis data
Usage
from datasets import… See the full description on the dataset page: https://huggingface.co/datasets/flodraye/autointerp-gpt2-multilingual-70.
