INDEX

Explanations

names of individuals

oai_token-act-pair · gpt-3.5-turbo

proper names, specifically related to individuals and their roles or titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ccording

-0.74

 deviations

-0.72

 entails

-0.67

 suppose

-0.67

products

-0.66

shows

-0.64

 glim

-0.64

 embodiments

-0.64

 necessities

-0.62

 differed

-0.62

POSITIVE LOGITS

Jr

1.43

III

1.04

Sr

1.03

JR

1.01

hetti

0.90

oglu

0.89

kson

0.89

Jr

0.87

otti

0.87

son

0.86

Activations Density 0.176%

names of individuals

proper names, specifically related to individuals and their roles or titles

No Comments

No Known Activations