INDEX

Explanations

The neuron searches for mentions of specific wars

oai_token-act-pair · gpt-3.5-turbo

references to war-related topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

FORMATION

-0.86

essee

-0.79

Ħ¢

-0.79

Asset

-0.78

TERN

-0.74

AUT

-0.72

sembly

-0.71

Distance

-0.71

Magikarp

-0.69

ostics

-0.68

POSITIVE LOGITS

fare

1.14

lords

1.10

lord

1.08

riors

1.06

rior

1.04

ring

0.98

 waged

0.96

fighting

0.91

fighter

0.89

war

0.86

Activations Density 0.032%

The neuron searches for mentions of specific wars

references to war-related topics

No Comments

No Known Activations