INDEX

Explanations

names or titles of individuals with a specific job role or rank, such as directors

oai_token-act-pair · gpt-3.5-turbo

instances of the word "director" with varying capitalization

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

 curve

-0.62

theless

-0.62

ulz

-0.62

asions

-0.61

 Leafs

-0.60

 conver

-0.59

adesh

-0.59

 skirm

-0.59

ãĥ¼ãĥ³

-0.58

Beg

-0.57

POSITIVE LOGITS

ial

1.05

ially

1.04

ials

0.92

IAL

0.91

director

0.89

Director

0.83

ovie

0.81

itatively

0.81

ate

0.77

nered

0.77

Activations Density 0.029%

names or titles of individuals with a specific job role or rank, such as directors

instances of the word "director" with varying capitalization

No Comments

No Known Activations