INDEX

Explanations

adjectives describing extremes, often negative

oai_token-act-pair · gpt-3.5-turbo

instances of the word "too" indicating excessiveness or intensity

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ords

-0.79

-0.77

hyde

-0.70

enance

-0.69

riots

-0.69

inarily

-0.68

craft

-0.68

creator

-0.67

anwhile

-0.65

issance

-0.65

POSITIVE LOGITS

 risky

0.82

far

0.79

 tempting

0.77

 busy

0.77

 costly

0.77

 distracting

0.76

 simplistic

0.76

 afraid

0.75

 bulky

0.74

 frail

0.74

Activations Density 0.037%

adjectives describing extremes, often negative

instances of the word "too" indicating excessiveness or intensity

No Comments

No Known Activations