INDEX

Explanations

proper nouns related to specific locations or names with the pattern "Id something"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Id" followed by numbers

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

UNCH

-0.84

 guiActiveUnfocused

-0.84

ICAN

-0.77

ABLE

-0.73

efully

-0.68

Beg

-0.68

ILA

-0.68

chnology

-0.67

ãģ¦

-0.66

ITCH

-0.65

POSITIVE LOGITS

iots

1.28

entity

1.17

aho

1.12

iom

1.11

ividual

1.09

iosyncr

1.08

enna

0.98

leness

0.93

gew

0.90

unn

0.89

Activations Density 0.009%

proper nouns related to specific locations or names with the pattern "Id something"

instances of the word "Id" followed by numbers

No Comments

No Known Activations