INDEX

Explanations

adjectives describing an extreme negative evaluation of something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "too" expressing excessiveness or overwhelming qualities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

-0.87

ords

-0.73

inarily

-0.71

issance

-0.68

anwhile

-0.67

riots

-0.66

here

-0.65

craft

-0.64

enance

-0.63

intel

-0.63

POSITIVE LOGITS

 busy

0.89

far

0.88

 tempting

0.85

 distracting

0.85

 afraid

0.84

 much

0.84

 simplistic

0.83

 risky

0.82

 costly

0.80

 frail

0.80

Activations Density 0.041%

adjectives describing an extreme negative evaluation of something

instances of the word "too" expressing excessiveness or overwhelming qualities

No Comments

No Known Activations