INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

Repeat

-0.70

å§«

-0.69

 antagonists

-0.62

FY

-0.62

Via

-0.62

 Defenders

-0.62

 Demand

-0.59

Secure

-0.58

 Guru

-0.57

ADD

-0.57

POSITIVE LOGITS

0.97

etary

0.77

anguage

0.77

tones

0.75

lement

0.75

etr

0.73

llah

0.71

mosp

0.71

hyde

0.71

 Awakens

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.