INDEX

Explanations

phrases indicating positive qualities or actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "good."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

eds

-0.79

idon

-0.76

eters

-0.72

anwhile

-0.72

ifles

-0.71

chan

-0.67

lees

-0.66

hani

-0.66

arthed

-0.66

osures

-0.66

POSITIVE LOGITS

 chunk

1.15

enough

1.13

sword

0.99

 approximation

0.99

 enough

0.94

 deal

0.92

ol

0.91

 idea

0.88

 amount

0.87

 example

0.86

Activations Density 0.068%

phrases indicating positive qualities or actions

instances of the word "good."

No Comments

No Known Activations