INDEX

Explanations

phrases related to a specific entity or brand name (e.g., "Boxer")

oai_token-act-pair · gpt-3.5-turbo

references to the term "Box" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ittee

-0.78

opoulos

-0.76

nee

-0.74

puter

-0.73

uman

-0.70

esthetic

-0.70

merce

-0.70

manent

-0.69

odic

-0.69

maj

-0.68

POSITIVE LOGITS

Box

1.36

Box

1.24

box

1.04

boxes

0.94

wra

0.87

box

0.86

es

0.86

 boxes

0.85

BOX

0.85

er

0.83

Activations Density 0.005%

phrases related to a specific entity or brand name (e.g., "Boxer")

references to the term "Box" in various contexts

No Comments

No Known Activations