INDEX

Explanations

proper nouns, specifically names of people and places

oai_token-act-pair · gpt-3.5-turbo

names or terms related to individuals and places

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

estinal

-0.73

fit

-0.68

holes

-0.67

manship

-0.67

imity

-0.65

 submar

-0.65

enger

-0.63

marks

-0.62

 ultr

-0.61

clair

-0.60

POSITIVE LOGITS

hyde

0.94

cade

0.90

ãĥķãĤ¡

0.75

ĳå£«

0.75

icles

0.72

allery

0.71

INGTON

0.71

ãĥ¼ãĥĨãĤ£

0.69

onel

0.69

mson

0.67

Activations Density 0.049%

proper nouns, specifically names of people and places

names or terms related to individuals and places

No Comments

No Known Activations