INDEX

Explanations

mentions of the word "te" with a high level of activation

oai_token-act-pair · gpt-3.5-turbo

instances of the word "te."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 eleph

-0.71

ipolar

-0.65

interrupted

-0.65

 justified

-0.63

 continents

-0.62

lessly

-0.61

etheless

-0.61

ĸļ

-0.61

 pressures

-0.61

 fronts

-0.60

POSITIVE LOGITS

brate

1.42

brates

1.30

ller

1.20

achers

1.12

levision

1.10

llers

1.08

achable

1.06

legram

1.04

legraph

1.03

aching

1.02

Activations Density 0.017%

mentions of the word "te" with a high level of activation

instances of the word "te."

No Comments

No Known Activations