INDEX

Explanations

mentions of sending or willingness to send people, troops, forces, or messages

oai_token-act-pair · gpt-3.5-turbo

instances of the word "send" and its variations in context of military or aid-related actions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Times

-0.64

urity

-0.63

minster

-0.63

phe

-0.62

Fram

-0.62

 Lange

-0.62

OWER

-0.62

Cola

-0.59

ESS

-0.59

theless

-0.58

POSITIVE LOGITS

inel

0.96

imental

0.90

entious

0.90

ãĥĥãĥī

0.85

keys

0.85

iments

0.83

maid

0.82

enced

0.82

itional

0.78

enc

0.74

Activations Density 0.028%

mentions of sending or willingness to send people, troops, forces, or messages

instances of the word "send" and its variations in context of military or aid-related actions

No Comments

No Known Activations