INDEX

Explanations

names or abbreviations containing "ne."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "ne"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

rador

-0.86

inarily

-0.86

hips

-0.83

allery

-0.77

 glim

-0.76

enhagen

-0.74

rament

-0.74

ienced

-0.74

awaru

-0.73

redited

-0.73

POSITIVE LOGITS

jad

0.97

arest

0.95

gger

0.94

cht

0.93

cker

0.92

phrine

0.92

quin

0.91

castle

0.89

zel

0.89

cks

0.87

Activations Density 0.031%

names or abbreviations containing "ne."

instances of the word "ne"

No Comments

No Known Activations