INDEX

Explanations

names or references to a person named "Elle"

oai_token-act-pair · gpt-3.5-turbo

mentions of prominent individuals or entities

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 crumbling

-0.66

 Icelandic

-0.64

razil

-0.64

yrinth

-0.64

itious

-0.64

ļé

-0.64

ritic

-0.63

keye

-0.62

irements

-0.62

 Scandinavian

-0.61

POSITIVE LOGITS

tto

1.09

ttes

0.91

elle

0.82

tta

0.77

izabeth

0.77

oad

0.76

issance

0.75

tti

0.75

lements

0.73

lement

0.73

Activations Density 0.018%

names or references to a person named "Elle"

mentions of prominent individuals or entities

No Known Activations