INDEX

Explanations

names of specific locations or individuals

oai_token-act-pair · gpt-3.5-turbo

variations of the word "able."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ORGE

-0.86

eve

-0.84

xon

-0.74

IDA

-0.72

meal

-0.69

 Catalyst

-0.67

lli

-0.66

 Mour

-0.66

women

-0.65

birth

-0.64

POSITIVE LOGITS

anca

1.20

ocal

0.99

abl

0.98

eness

0.96

osure

0.92

anches

0.90

ingu

0.89

acas

0.82

anty

0.82

ocaly

0.80

Activations Density 0.009%

names of specific locations or individuals

variations of the word "able."

No Comments

No Known Activations