INDEX

Explanations

references to the city of Waterloo or Wellington

oai_token-act-pair · gpt-3.5-turbo

the occurrences of the terms "Waterloo" and "Wellington"

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

ures

-0.85

manship

-0.80

ibli

-0.78

ively

-0.77

 charact

-0.76

lishing

-0.74

nar

-0.74

ãģ¦

-0.74

es

-0.73

MENTS

-0.71

POSITIVE LOGITS

 Waterloo

0.92

yip

0.83

ategory

0.83

 Denis

0.82

 Laur

0.82

agine

0.80

 Wellington

0.79

renheit

0.79

oleon

0.78

ipop

0.76

Activations Density 0.024%

references to the city of Waterloo or Wellington

the occurrences of the terms "Waterloo" and "Wellington"

No Comments

No Known Activations