INDEX

Explanations

terms related to indirect effects or actions

oai_token-act-pair · gpt-3.5-turbo

references to indirect relationships or actions

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

ikers

-0.75

Bre

-0.74

 Disorder

-0.73

uster

-0.72

mbuds

-0.69

 Archdemon

-0.68

nesota

-0.67

mberg

-0.67

 Beware

-0.66

topia

-0.66

POSITIVE LOGITS

 indirect

1.28

irect

1.01

 indirectly

0.82

yrinth

0.80

 intermediary

0.76

 illumination

0.76

repre

0.73

 prox

0.73

 causal

0.72

 sunlight

0.72

Activations Density 0.010%

terms related to indirect effects or actions

references to indirect relationships or actions

No Comments

No Known Activations