INDEX

Explanations

It seems there is no clear pattern of activation for Neuron 4 as it does not activate for any of the provided tokens. Without any non-zero activations to analyze, we cannot determine what this neuron is looking for

oai_token-act-pair · gpt-4-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cou

-0.77

 Raider

-0.70

 Hunters

-0.70

 merc

-0.64

 Kardash

-0.64

 recess

-0.63

rogen

-0.63

obin

-0.63

heid

-0.59

yrinth

-0.58

POSITIVE LOGITS

Offline

0.71

Best

0.69

 Cosponsors

0.69

 RELE

0.67

)</

0.66

mob

0.65

nces

0.64

affe

0.62

airo

0.61

Operation

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.