INDEX

Explanations

phrases related to effects or impacts

oai_token-act-pair · gpt-3.5-turbo

references to the "Effect" phenomenon 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 chau

-0.80

odan

-0.67

yahoo

-0.67

 grab

-0.66

itialized

-0.66

 apologize

-0.64

url

-0.64

urous

-0.63

 pant

-0.62

 customary

-0.62

POSITIVE LOGITS

 Effect

2.96

effect

2.45

Effect

2.15

effects

1.41

Effects

1.40

 Effects

1.35

inkle

1.21

EFF

1.11

 effectiveness

1.09

EFF

1.04

Activations Density 0.025%

phrases related to effects or impacts

references to the "Effect" phenomenon in various contexts

No Comments

No Known Activations